1 ;;;; -*- coding: utf-8; -*-
2 changes in sbcl-0.9.5 relative to sbcl-0.9.4:
3 * new feature: timers based on Zach Beane's excellent timer package
4 * added support for the following external formats: koi8-u,
5 x-mac-cyrillic, cp437, cp850, cp852, cp855, cp857, cp860, cp861,
6 cp862, cp863, cp864, cp865, cp866, cp869, cp874, iso-8859-2,
7 iso-8859-3, iso-8859-4, iso-8859-5, iso-8859-6, iso-8859-7,
8 iso-8859-8, iso-8859-9, iso-8859-10, iso-8859-11, iso-8859-13,
9 iso-8859-14, cp1250, cp1251, cp1252, cp1253, cp1254,
10 cp1255,cp1256, cp1257, cp1258 (contributed by Ivan Boldyrev)
11 * incompatible change: a threaded SBCL will no longer revert to
12 non-threaded mode on non-NPTL systems, but refuse to start entirely.
13 * bug fix: interrupts are disabled until startup is complete; no
14 more sigsegvs when receiving a signal to soon
15 * optimization: faster 32-bit SB-ROTATE-BYTE:ROTATE-BYTE on non-x86/ppc
17 * bug fix: add a workaround for the memory randomization features in
18 Linux kernels >= 2.6.12 that interfere with SBCL's memory maps. This
19 workaround will only be in effect on systems with the proc filesystem
21 * bug fix: printing objects of type HASH-TABLE signals a
22 PRINT-NOT-READABLE error when *READ-EVAL* is NIL. (reported by
24 * bug fix: GET-INTERNAL-REAL-TIME now works even for processes that
25 have been running for over 50 days. (reported by Gilbert Baumann)
26 * bug fix: the logic for getting names of functions gets less
27 confused when confronded with alternate-metaclass
28 funcallable-instances. (reported by Cyrus Harmon)
29 * bug fix: FUNCTIONP and (LAMBDA (X) (TYPEP X 'FUNCTION)) are now
30 consistent, even on internal alternate-metaclass objects.
31 * bug fix: SB-MOP:FUNCALLABLE-STANDARD-OBJECT is now a subclass of
32 STANDARD-OBJECT, as required by AMOP.
33 * bug fix: the classes STANDARD-CLASS and
34 SB-MOP:FUNCALLABLE-STANDARD-CLASS are now compatible in the
35 SB-MOP:VALIDATE-SUPERCLASS sense; there remains a constraint about
36 finalized classes and the FUNCTION class.
37 * bug fix: the SB-MOP:METAOBJECT class is now implemented as
39 * bug fix: flush closure information collected by physical
40 environment analysis prepass before the main pass. (bug reported
42 * bug fix: compiler pack phase does not modify a hashtable, which is
43 iterated. (reported by Bryan O'Connor, fixed by Rob MacLachlan)
44 * bug fix: backquote does not optimize (LIST* x y) to CONS when x
45 may be expanded. (reported by Alexander <malishev> on c.l.l)
47 ** bug fix: parent thread now can be gc'ed even with a live
49 ** bug fix: binding a special with PROGV to no value is not
50 the same as that symbol not having been bound (thanks to
52 ** bug fix: binding specials is thread safe (thanks to
54 ** bug fix: interrupt handlers are now per-process, RUN-PROGRAM
55 and SB-SPROF do not die with 'no handler for signal XX in
56 interrupt_handle_now(..)' anymore
57 ** bug fix: WITH-TIMEOUT works with multiple running threads
58 * fixed some bugs revealed by Paul Dietz' test suite:
59 ** ENSURE-GENERIC-FUNCTION should take a method class object for
60 the :method-class keyword argument.
62 changes in sbcl-0.9.4 relative to sbcl-0.9.3:
63 * new port: the Solaris operating system on x86 processors is now
64 mostly supported, though some rough edges in the environment
65 remain. (thanks to Daisuke Homma)
66 * enhancement: SBCL on MIPS platforms now has a much larger dynamic
67 space for its heap. (thanks to Thiemo Seufer)
68 * optimization: SBCL on MIPS platforms now supports dynamic-extent
69 closures. (thanks to Thiemo Seufer)
70 * minor incompatible change: eof selects abort in the debugger.
71 * minor incompatible change: *INVOKE-DEBUGGER-HOOK* is run before
72 *DEBUGGER-HOOK* => *DEBUGGER-HOOK* is not run when the debugger
74 * minor incompatible change: SB-KERNEL:INSTANCE-LAMBDA is
75 deprecated, and will go away in a future revision of SBCL.
76 * minor incompatible change: GC-ON and GC-OFF are no longer
77 implemented with a counter, it does not matter how many times gc
79 * bug fix: discriminating functions for generic function classes
80 with non-standard methods for COMPUTE-APPLICABLE-METHODS no longer
81 make invalid assumptions about method precedence order. (reported
83 * bug fix: TRUNCATE now correctly signals division by zero on MIPS
84 platforms. (thanks to Thiemo Seufer)
85 * bug fix: degree sign (<U00B0>) could not be encoded in KOI8-R.
86 * bug fix: correct pathname printing with printer escaping is on.
87 (thanks to Kevin Reid)
88 * bug fix: complex VOP definitions in "user-space" no longer trigger
89 package locks. (reported by Zach Beane)
90 * fixed bug 343: SB-KERNEL:INSTANCE-LAMBDA is no longer necessary
91 for funcallable-instance functions, and is no different from
93 * bug fix: PARSE-INTEGER no longer depends on the whitespaceness of
94 characters in the current readtable. (reported by Nicholas Neuss)
95 * bug fix: SUBTYPEP on various CONS types returns more a more
96 accurate acknowledgment of its certainty.
97 * optimizations: REMOVE-DUPLICATES now runs in linear time on
98 lists in some cases. This partially fixes bug 384.
99 * flush all standard streams before prompting in the REPL and the
101 * bug fix: signal handling and triggering gc do not conflict
102 directly anymore, in particular a high frequency sb-sprof does
103 not prevent gc from running
104 * bug fix: DECODE-UNIVERSAL-TIME now uses a more reasonable
105 approximation for timezone and DST information between the
106 universal time epoch and the smallest negative 32-bit time_t.
107 * bug fix: ENCODE-UNIVERSAL-TIME no longer errors when passed the
108 year 1899 so long as the encoded time is non-negative.
109 * bug fix: on the x86 platform, SB-ROTATE-BYTE:ROTATE-BYTE returns
110 the correct answer for non-constant rotate amounts when performing
111 an inline 32-bit rotation.
113 ** bug fix: RELEASE-FOREGROUND doesn't choke on session lock if
114 there is only one thread in the session
115 ** bug fix: memory leak for streams created in one thread and
116 written to in another
117 ** bug fix: lockup when compiled with gcc4
118 ** bug fix: race that allows the gc to be triggered when gc is
120 ** bug fix: one less memory fault in INTERRUPT-THREAD, again
121 ** bug fix: gc and INTERRUPT-THREAD don't hang when the RT signal
123 ** bug fix: finished threads are gc'ed properly
124 * fixed some bugs revealed by Paul Dietz' test suite:
125 ** CALL-NEXT-METHOD signals an error (in safe code) when the call
126 has arguments with a different set of applicable methods from
127 the orignal arguments.
128 ** The type error thrown by MAP now has a correct expected-type
130 ** DISASSEMBLE now throws a TYPE-ERROR when its argument does not
131 name a compiled function.
132 ** Three MISC tests where a large bignum was improperly coerced to
133 a float that couldn't represent that bignum during type
134 derivation were fixed.
135 ** SUBTYPEP can now handle types involving the negation of a
136 list-form FUNCTION type.
137 ** SUBTYPEP also now handles certain COMPLEX type specifiers such
138 as (COMPLEX (AND RATIO (NOT FIXNUM))).
139 ** READ-BYTE and WRITE-BYTE no longer take stream designators.
141 changes in sbcl-0.9.3 relative to sbcl-0.9.2:
142 * New feature: Experimental support for bivalent streams: streams
143 opened with :element-type :default now allow character and binary
144 (unsigned-byte 8) I/O
145 * Support for the koi8-r external format. (thanks to Ivan Boldyrev)
146 * Bug fix: OPEN no longer fails when *PRINT-READABLY* is T. (thanks
148 * bug fix: no more highly sporadic "couldn't check whether ~S is
149 readable" when reading a stream and an interrupt hits in the middle
150 of a select system call
151 * compiler better recognizes complex arrays (reported by Eduardo
153 * bug fix: out-of-line SB-SYS:FOREIGN-SYMBOL-ADDRESS did not work
154 for variables on SBCL built with linkage-tables. (reported by Luis
156 * various error reporting improvements.
157 * optimizations: LOGNOR on fixnums is improved in the MIPS backend.
158 (Thanks to Thiemo Seufer)
159 * bug fix: nested reader invokations work correctly
160 * bug fix: it is possible to have simultaneous references to foreign
161 code and foreign data with the same name.
163 ** added x86-64 support
164 ** incompatible change: the threading api now works with thread
165 objects instead of thread ids
166 ** bug fix: threads are protected from signals and interruption when
167 starting up or going down
168 ** bug fix: a race where an exiting thread could lose its stack to gc
169 ** bug fix: don't halt on infinite error in threads if possible
170 ** fixed numerous gc deadlocks introduced in the pthread merge
171 ** bug fix: fixed thread safety issues in read and print
172 ** bug fix: debugger doesn't hang on session lock if interrupted at
173 an inappropriate moment
174 ** bug fix: run-program is now thread safe(r)
175 ** bug fix: inner with-recursive-lock no longer releases the mutex
176 * fixed a bug in (DECODE-UNIVERSAL-TIME 0) (reported by Paul Dietz)
177 * fixed some bugs revealed by Paul Dietz' test suite:
178 ** TYPE-ERRORs from signalled by COERCE now have DATUM and
179 EXPECTED-TYPE slots filled.
180 ** MULTIPLE-VALUE-SETQ always returns the primary value of the
182 ** MAKE-CONDITION accepts classes as type-designators.
183 ** COMPILE may never return NIL.
184 ** ENCODE-UNIVERSAL-TIME now guards against times out of Unix's
185 range before calling Unix time functions
187 changes in sbcl-0.9.2 relative to sbcl-0.9.1:
188 * numerous signal handling fixes to increase stability
189 * Support for EUC-JP external format. (thanks to NIIMI Satoshi)
190 * minor incompatible change: we now correctly canonize default
191 initargs, making them be a list of (INITARG INITFORM INITFUNCTION)
192 as per the MOP, rather than the historical (INITARG INITFUNCTION
193 INITFORM). (reported by Bruno Haible)
194 * new feature: WITH-COMPILATION-UNIT now accepts a non-standard
195 :SOURCE-PLIST option. See (DOCUMENTATION #'WITH-COMPILATION-UNIT T)
196 for more information.
197 * TRUENAME and PROBE-FILE now correctly resolve symlinks even if the
198 pathname is a directory pathname.
199 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
200 * dynamic space size on PPC has been increased to 768Mb. (thanks to
202 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
203 :METACLASS argument in addition to a class name. (reported by
204 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellmann)
205 * RESTART-CASE can now be wrapped around CALL-METHOD forms.
206 (reported by Bruno Haible; patch from Gerd Moellmann)
207 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
209 * bug fix: more cleanups to the floating point exception handling on
210 x86-64 (thanks to James Knight)
211 * bug fix: the compiler does not try to constant fold calls of
212 COERCE and BIT-* functions when they return freshly constructed
213 objects. (reported by Nikodemus Siivola and Paul Dietz)
214 * optimization: improved the MIPS versions of generic (in the
215 generic sense) arithmetic routines. (thanks to Thiemo Seufer)
216 * optimization: direct conversion of (unsigned-byte 32) to floats on
217 the PowerPC platform.
218 * optimization: structure instances with raw slots now use less
219 memory, and probably show better memory locality. (thanks to
221 * optimization: DYNAMIC-EXTENT declarations for lists and closures
222 are treated as requests for stack allocation on the x86-64,
223 Alpha-32, PPC and SPARC platforms.
224 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
225 the wrong file descriptor; implementation of SOCKET-OPEN-P.
226 (thanks to Tony Martinez)
228 ** gcing a dead thread can no longer lead to lockups
229 ** threads block signals until they are set up properly
230 ** errno is no longer shared by threads
231 ** interrupt-thread restores the eflags register on x86
232 ** fixed some lockups due to gc/thread interaction
233 * fixed some bugs revealed by Paul Dietz' test suite:
234 ** invalid dotted lists no longer raise a read error when
236 ** signal an error if a symbol that names a declaration is used
237 as the name of a type, or vice versa
238 ** allow using the (declare (typespec var*)) abbreviation for
239 (declare (type typespec var*)) with all type specifiers
240 ** circularity detection works properly with PPRINT-LOGICAL-BLOCK
241 ** always return NIL from PPRINT-POP when OBJECT is NIL
242 ** don't signal errors when pretty-printing malformed LABELS,
243 FLET or MACROLET forms
244 ** declarations in a DOLIST body are also in scope for the
246 ** COMPILE-FILE accepts all pathname designators as INPUT-FILE
248 ** the ELEMENT-TYPE keyword argument to WITH-OUTPUT-STREAM is
251 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
252 * fixed cross-compiler leakages that prevented building a 32-bit
253 target with a 64-bit host compiler.
254 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
255 opened with :IF-EXISTS :OVERWRITE.
256 * fixed bug 281: error for an invalid qualifier in a short-form method
257 combination method is not signalled until the faulty method is called.
258 * bug fix: iteration variable type inferrer failed to deal with open
259 intervals. (reported by Alan Shields)
260 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
261 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
262 or more runtime options were provided to the sbcl binary.
263 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
264 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
265 * optimization: there's now a fast-path for fixnum arguments in the
266 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
267 * optimization: the code generated on x86-64 is more compact thanks
268 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
269 * bug fix: floating-point exception handling now partly works on
270 x86-64. (Thanks to James Knight)
271 * improvement to the MIPS backend from Thiemo Seufer: C-style
272 64-bit long long arguments and return values to alien functions
274 * fixed some bugs revealed by Paul Dietz' test suite:
275 ** the type-error signalled from WARN has a filled-in DATUM slot.
276 ** the type-error required when a stream is not associated with
277 a file has the stream as its datum.
278 ** type-errors on single-floats on x86-64 no longer have
279 :INVALID-OBJECT as the datum
280 ** the type-errors signalled for invalid function names now have
281 a correct expected type
282 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
283 for structures defined with a :TYPE. Documentation strings for
284 typed structures are no longer immediately discarded
285 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
286 broadcast streams and synonym streams. FILE-LENGTH now also works
287 on broadcast streams.
289 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
290 * incompatible change: the --noprogrammer option, deprecated since
291 version 0.7.5, has been removed. Please use the equivalent
292 --disable-debugger option instead.
293 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
295 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
296 inoperational for a while now) has been completely removed.
297 * null lexical environments are now printed as #<NULL-LEXENV>,
298 significantly reducing the amount of clutter in typical backtraces.
299 * documentation on weak pointers, finalization, and after GC hooks
300 has been added to the manual.
301 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
302 as other specialized array types, is much faster. SUBSEQ and
303 COPY-SEQ on such arrays have also been sped up.
304 * optimization: EQL is now more efficient when at least other argument
305 is known to be of type (OR FIXNUM (NOT NUMBER)).
306 * fixed bug: the runtime is now less vulnerable to changes in the
307 size of the SBCL object on OS X, and virtual memory is reserved for
308 all spaces that need to be at a fixed address.
309 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
310 * fixed bug: finalizers and after GC hooks that cause consing are now
312 * fixed bug: compiler error messages and summaries are now printed to
313 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
314 * fixed inference of the upper bound of an iteration variable.
315 (reported by Rajat Datta).
316 * fixed bug 211e: calling local functions with duplicated constant
317 keyword argument no longer causes a bogus style warning about an
319 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
320 ftype declaration to be disregarded. (reported by Dave Roberts)
321 * fixed bug 373: caused by erronous compilation of references to alien
322 variables in the runtime on ppc/darwin.
323 * fixed bug 376: CONJUGATE type deriver.
324 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
325 (reported by Baughn on #lisp)
326 * a cross-compiler bug on non-x86 platforms has been identified and
327 fixed. (thanks to Bruno Haible)
328 * improvements to the MIPS runtime code for increased stability.
329 (thanks to Thiemo Seufer)
330 * increased the maximimum compact environment size to allow
331 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
332 * partial workaround for bug 135: don't name the function we're
333 calling for hairy cases of EVAL, so as not to accumulate one environment
334 entry per call to EVAL. (reported by Kevin Reid)
335 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
336 * optimization: type testing for non-vector arrays should be faster.
337 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
339 * bug fix: setting 31st element of a bit vector to zero did not work
341 * bug fix: redefining a class definition which failed due to a
342 previous accessor / function clash now works (but see BUGS entry
343 #380 for more problems in this area). (thanks to Zach Beane)
344 * the long form of DEFINE-METHOD-COMBINATION disables method group
345 checking when given a single method group with pattern *. (thanks
347 * on x86 compiler supports stack allocation of results of simple
348 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
349 * support for building and running on Mac OS X 10.4 "Tiger" has been added
350 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
351 * fixed some bugs related to Unicode integration:
352 ** the restarts for recovering from input and output encoding
353 errors only appear when there is in fact such an error to
355 * fixed some bugs revealed by Paul Dietz' test suite:
356 ** MISC.549 and similar: late transformation of unsafe type
357 assertions into derived types caused unexpected code
359 ** SCALE-FLOAT type deriver is less wrong.
360 ** type derivers for EXP, LOG and similar functions compute result
361 types for complex arguments better.
362 ** (MISC.563) CONJUGATE type deriver works for very restricted
364 ** out-of-line type testers for character strings are available.
365 ** EQUAL compiler transform understands specialness of objects
367 ** accessing double-floats stored on the stack now works on x86-64.
368 ** debugger internals could sometimes create invalid lispobjs,
369 resulting in GC crashes.
370 ** MISC.548: type check weakening can convert required type into
372 ** initialization forms for bindings are not in scope of free special
375 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
376 * incompatible change: thread support for non-NPTL systems has
377 been removed - locking is buggy and unreliable. A threaded
378 SBCL build will now warn at startup and refuse to create
379 new threads, unless futex support is detected in the kernel
380 * incompatible change: the top level REPL now has only an ABORT
381 restart associated with it, not TOPLEVEL and ABORT as it used to.
382 TOP and TOPLEVEL are now available as debugger commands for
383 returning to the top level.
384 * incompatible change: forms evaluated in the REPL now use the
385 global optimization policy.
386 * incompatible change: user- and system-initialization files are
387 no longer processed with LOAD, but by READ and EVAL; hence the
388 global optimization policy, startup package, readtable, etc,
390 * The .fasl file format number has been incremented because of
391 various incompatible changes.
392 * internal entry point details and argument counts no longer appear
393 in backtraces unless explicitly requested by setting
394 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
395 * built-in and standard functions no longer have names like "top
396 level local call to FOO".
397 * fixed bug 32: functions defined in non-null lexical environments
398 now have more legible printed representation
399 * fixed bug 33: functions defined in non-null lexical environemnts
400 are now more amenable to inspection by INSPECT.
401 * workaround for bug 354: XEPs no longer appear in backtraces unless
402 explicitly requested.
403 * fixed bug: receiving the signal which results from attempting to
404 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
405 SIGBUS on other *BSDs) on architectures where the C stack is also
406 the Lisp stack (x86 and x86-64) from foreign code no longer leads
407 to debugger confusion or wild instability. (reported by Cheuksan
409 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
410 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
411 (reported by Lutz Euler)
412 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
413 compiled code. (reported by Paul Dietz)
414 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
415 specialized parameter to a method to something that is not TYPEP
416 the specializer is now possible.
417 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
418 face of package deletion.
419 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
420 pathname. (reported by David Tolpin on c.l.l)
421 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
422 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
423 STRING-STREAM and FILE-STREAM, respectively.
424 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
425 than previously; in addition, its test suite is now run on build.
426 (thanks to Luke Gorrie)
427 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
428 * optimization: added a immediate representation for single-floats
430 * fixed some bugs revealed by Paul Dietz' test suite:
431 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
432 ** attempting to create a package with a colliding nickname causes
433 correctable errors to be signalled.
434 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
435 ** several x86-64 backend bugs related to sign-extension of immediate
438 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
439 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
440 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
442 * fixed loading of multiply forward-referenced layouts.
443 (thanks to Cheuksan Wang)
444 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
445 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
446 (see COMPILE-FILE documentation for details of :PRINT option)
447 * fixed bugs 19 and 317: fixed-format floating point printing is
448 more accurate. This also fixes a bug reported by Adam Warner
449 related to the ~@F format directive.
450 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
452 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
453 dispatch table if the from-char is a dispatch macro character.
454 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
455 by the wordsize no longer ignore the last word. (reported by Lutz
457 * fixed bug in type checking of dynamic-extent variables. (reported
459 * optimization: sequence traversal functions use their freedom to
460 coerce function designators to functions.
461 * optimization: code with many calls to CLOS methods specialized on
462 CLOS classes has had redundant type checks removed.
463 * optimization: added declarations to speed up operations that access
464 the internal character database (for example STRING-UPCASE)
465 * optimization: comparison operations between floats and sufficiently small
466 fixnums no longer create extra rationals
467 * fixed some bugs related to Unicode integration:
468 ** portions of multibyte characters at the end of buffers for
469 character-based file input are correctly transferred to the
470 start of the buffer at the next read.
471 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
472 passing it through to OPEN.
473 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
474 argument given to internal calls to OPEN.
475 ** the built-in comment readers (introduced by character sequences
476 ";" and "#|") are more forgiving to encoding errors; they will
477 STYLE-WARN and then attempt to resync the stream at a character
478 boundary. (thanks to Teemu Kalvas)
479 * fixed some bugs in the x86-64 port:
480 ** Negative short int return values from c-calls are sign-extended
482 ** The stack is aligned to 16-bytes for c-calls, as required by
483 the ABI. (thanks to Cheuksan Wang)
484 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
485 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
486 for enums. (thanks to Vincent Arkesteijn)
487 ** Multiple small optimizations and bugfixes for floating point
489 * fixed some bugs revealed by Paul Dietz' test suite:
490 ** Space, Tab, Linefeed, Return and Page have the invalid
491 secondary constituent character trait.
492 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
494 ** WITH-INPUT-FROM-STRING should only update the index place on
496 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
498 ** Bit-array operations (BIT-AND and similar) worked incorrectly
499 with one-dimensional arrays with fill pointers.
500 ** TYPE-OF failed on a complex with an integer realpart and a
502 ** compiler failure during type inference for the code of form
503 (IF (EQL X (THE ...)) ...) (MISC.535).
505 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
506 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
507 platform. The port supports 61-bit fixnums, large memory spaces
508 and reloading shared object files.
509 * enhancement: saving cores with foreign code loaded is now
510 supported on ppc/Darwin in addition to the previously supported
512 * enhancement: the statistical profiler now walks deeper into the
513 call stack for more meaningful call-graphs and accrued time
514 reports (x86/x86-64 only). It also now reports time spent in
516 * enhancement: it is now possible to trace most individual methods
517 of a generic function in addition to tracing the generic function
519 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
521 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
522 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
523 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
524 produces an error. (thanks to Vincent Arkesteijn)
525 * bug fix: NAMESTRING on pathnames with :WILD components in their
526 directories works correctly. (thanks to Artem V. Andreev)
527 * fixed bug 125: compiler preserves identity of closures. (reported
529 * bug fixed: functions with &REST arguments sometimes failed with
530 "Undefined function" when compiled with (DEBUG 3). (reported by
532 * bug fix: overflow during compiling of setting element of a bit
533 vector with constant index and value. (reported by Timmy Douglas)
534 * build fix: fixed the dependence on *LOAD-PATHNAME* and
535 *COMPILE-FILE-PATHNAME* being absolute pathnames.
536 * on x86 compiler partially supports stack allocation of dynamic-extent
538 * GO and RETURN-FROM do not check the extent of their exit points
539 when compiled with SAFETY 0.
540 * fixed some bugs related to Unicode integration:
541 ** encoding and decoding errors are now much more robustly
542 handled; it should now be possible to recover even from invalid
543 input or output to the terminal. (thanks to Teemu Kalvas)
544 ** provided a first cut at implementing STRING-TO-OCTETS and
545 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
546 ** altered the SB-MD5 contributed module slightly, changing the
547 interface just enough for it to be supportable for builds where
548 lisp characters are not eight bits.
549 * fixed some bugs revealed by Paul Dietz' test suite:
550 ** the FORMATTER-generated functions for ~V[ conditionals require
551 the correct number of arguments.
552 ** READ-FROM-STRING returns the mandated second value when applied
553 to displaced strings.
554 ** the #\Rubout and #\Backspace characters are treated as invalid
555 constituent characters by the tokenizer.
557 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
558 * new feature: reloading changed shared object files with
559 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
560 * new feature: references to foreign variables and functions
561 can now be compiled and loaded before the corresponding shared
562 object file is loaded, as long as the foreign definitions are
563 available at runtime.
564 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
565 supported. (thanks to Dan Debertin)
566 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
567 just keywords. (thanks to Vincent Arkesteijn)
568 * fixed bug #331: structure-class instances corresponding to
569 DEFSTRUCT forms are now created eagerly.
570 * fixed bug #345: backtraces from calls to undefined functions work
571 on x86 as well. Related bug #61 is now also partially fixed on x86
572 (backtraces from throws to unknown catch tags.)
573 * bug fix: lambda-list parsing is now stricter vrt. order and number
574 of lambda-list keywords.
575 * bug fix: as specified by AMOP, an error is signalled if a
576 class-option appears multiple times in a DEFCLASS form. (reported
578 * bug fix: RANDOM can be compiled when the compiler derives the type
579 of its numeric argument as a disjoint set of small integers.
580 (reported by Paul Dietz)
581 * bug fix: starting a core saved with shared objects loaded when
582 those objects are not available no longer causes threaded SBCL to
583 hang. (reported by Sean Ross)
584 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
586 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
587 argument, which must always be NIL. (reported by Kalle Niemitalo)
588 * bug fix: printing 1.0d+23 no longer results in an error.
589 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
590 * bug fix: structure slot setters preserve evaluation order. (thanks
592 * fixed some bugs related to Unicode integration:
593 ** RUN-PROGRAM can allow its child to take input from a Lisp
594 stream. (reported by Stefan Scholl)
595 ** ASDF-INSTALL successfully downloads .tar.gz files in all
596 locales. (reported by Ken Causey)
597 * fixed some bugs revealed by Paul Dietz' test suite:
598 ** INCF, DECF and REMF evaluate their place form as specified in
600 ** FORMATTER expands ~{ ~} iteration directives with V or #
601 parameters correctly.
602 ** FORMATTER deals with the ~@[ ~] conditional directive where the
603 consequent uses no arguments correctly.
604 ** the system has a partial understanding of the (COMPLEX RATIO)
606 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
609 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
610 * new feature: a build-time option (controlled by the :SB-UNICODE
611 keyword feature, enabled by default) for building the system with
612 support for the entire 21-bit character space defined by the
614 * new feature: the system now has rudimentary external-format
615 support; the primary user-visible change at this time is that
616 characters with the high bit set (such as lower-case-e-acute) will
617 print correctly to a terminal in a UTF-8 environment.
618 * minor incompatible change: BASE-CHAR no longer names a class;
619 however, CHARACTER continues to do so, as required by ANSI.
620 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
621 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
622 has been moved to the SB-EXT package (temporarily re-exported from
624 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
625 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
627 * fixed bug #351: better error-handlind and reporting for malformed
629 * fixed bug #350: bignum-printing is now more memory-efficient,
630 allowing printing of very large bignums, eg. (expt 2 10000000).
631 (reported by Bruno Haible)
632 * fixed bug #302: better primitive-type selection for intersection
634 * fixed bug #308: non-graphic characters now all have names, as
635 required. (reported by Bruno Haible)
636 * bug fix: redefining a class with different superclasses now correctly
637 removes it from the direct-subclasses of its previous superclasses.
638 (reported by David Morse)
639 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
640 new value now works. (reported by Bruno Haible)
641 * bug fix: correct canonicalization of multiple non-standard slot
642 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
643 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
644 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
645 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
646 now exists, an signals an error.
647 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
648 during macroexpansion for non-iteration variables as well. (reported
649 by Bruno Haible for CMUCL)
650 * bug fix: Cyclic structures and unprintable objects in compiler
651 messages no longer cause errors. (reported by Bruno Haible)
652 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
653 and READ-FROM-STRING all now return a primary value of NIL if
654 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
655 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
656 (reported by Bruno Haible for CMUCL)
657 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
658 do not designate a condition. (reported by Bruno Haible for
660 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
661 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
662 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
663 permitting resolution in favour of any of the conflicting symbols.
664 (reported by Bruno Haible for CMUCL)
665 * bug fix: EQUAL compiler optimizations is less aggressive on
666 strings which can potentially compare true despite having distinct
667 specialized array element types.
668 * bug fix: unit enumerations can be defined without dividing by
669 zero. (thanks to Vincent Arkesteijn)
670 * FORMAT compile-time argument count checking has been enhanced.
671 (report from Bruno Haible for CMUCL)
672 * a partial workaround for the bug 262: the compiler does not try to
673 inline-expand a local function doing RETURN-FROM from a deleted
674 BLOCK. (thanks to Peter Denno for the bug report and to David
675 Wragg for the simple test case)
676 * fixed some bugs revealed by Paul Dietz' test suite:
677 ** RENAME-PACKAGE allows all package designators as new package
679 ** constraint propagation and lambda variable substitution are
680 more cautious in dealing with partially deleted code.
681 ** compiler sometimes lost reoptimization passes.
682 ** CERROR, given a condition as condition designator, passes its
683 remaining arguments to the continue format control without
685 ** Case-altering FORMAT directives work correctly on non-ASCII
687 ** The REAL type specifier handles bounds outside the floating
688 point ranges without signalling FLOATING-POINT-OVERFLOW.
689 ** Functions with IR1-transformations can create intercomponent
690 references to global functions.
691 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
693 ** FORMAT ~R treats a nil value for its first parameter correctly.
695 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
696 * enhancement: saving cores with foreign code loaded is now
697 supported on x86/NetBSD and sparc/Linux in addition to the previously
699 * bug fix: on some platforms repeated installations caused multiple
700 copies of HTML documentation to be installed -- should not happen
701 any more. (reported by Stefan Scholl)
702 * bug fix: parsing self-recursive alien record types multiple times
703 no longer causes infinite recursion. (reported by Thomas F. Burdick,
704 original patch by Helmut Eller for CMUCL)
705 * bug fix: stack-exhaustion detection works now on NetBSD as well.
706 (thanks to Richard Kreuter)
707 * bug fix: defining classes whose accessors are methods on existing
708 generic functions in other (locked) packages no longer signals
709 bogus package lock violations. (reported by François-René Rideau)
710 * bug fix: special variables as DEFMETHOD parameters no longer have
711 associated bogus type declarations. (reported by David Wragg and
713 * bug fix: read-write consistency on streams of element-type
714 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
715 * bug fix: redefiniton of the only method of a generic function with
716 no DEFGENERIC no longer emits a full WARNING. In addition,
717 redefinition of generic functions with no DEFGENERIC to an
718 incompatible lambda list now signals an error. (thanks to Zach
720 * bug fix: DEFGENERIC now works even when there's a function of the
721 same name in an enclosing lexical environment. (thanks to Zach
723 * fixed compiler failure, caused by instrumenting code during
724 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
725 * optimization: added loop analysis and improved register allocation
726 * fixed some bugs revealed by Paul Dietz' test suite:
727 ** POSITION on displaced vectors with non-zero displacement
728 returns the right answer.
729 ** (SIMPLE-STRING) is a valid type specifier for sequence
731 ** *PRINT-LEVEL* handling for slotless structures is pedantically
733 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
734 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
736 ** The justification version of the FORMAT ~< directive treats
737 non-zero minpad parameter correctly.
739 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
740 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
741 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
742 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
743 the supported interface.
744 * new feature: Single-stepping of code compiled with DEBUG 2 or
745 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
746 * new feature: saving cores with foreign code loaded is now
747 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
748 Timothy Moore's work for CMUCL)
749 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
750 parameters to * instead of NIL if no initform is supplied.
751 (reported by Johan Bockgaard)
752 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
753 correctly when FUNCALL appears as the car of the form. Note:
754 despite this FUNCALL forms are not currently subject to
755 compiler-macro expansion. (port of Raymond Toy's fix for the
756 same from CMUCL, reported by Johan Bockgaard)
757 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
758 (thanks for Teemu Kalvas)
759 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
760 case instead of terminating the iteration (thanks for Julian
761 Squires, Sean Champ and Raymond Toy)
762 * bug fix: incorrect expansion of defgeneric that caused a style
763 warning. (thanks for Zach Beane)
764 * bug fix: slot accessor effective method computation works properly
765 for classes with multiple non-standard applicable methods on
766 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
767 * on x86 compiler supports stack allocation of results of LIST and
768 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
769 CMUCL implementation by Gerd Moellmann)
770 * fixed some bugs revealed by Paul Dietz' test suite:
771 ** FORMAT strings with both the ~<~:;~> form of the justification
772 directive and pretty-printing directives cause an error.
774 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
775 * incompatible change: the internal functions
776 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
777 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
778 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
779 instead of the old functions.
780 * new feature: on platforms where "dladdr" is available foreign
781 function names now appear in backtraces. (based on Helmut Eller's
783 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
784 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
786 * documentation: ASDF manual is now distributed with SBCL.
787 * bug fix: SBCL can now load its contributed modules with REQUIRE
788 even if the system-provided entries have been removed from
789 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
791 * bug fix: backtraces involving undefined functions or assembly
792 routines are more informative. (thanks to Brian Downing)
793 * bug fix: mutually referent alien structures now work correctly.
794 (reported by Rick Taube)
795 * bug fix: structures defined by WITH-ALIEN can be referred to
796 within other definitions in the same WITH-ALIEN.
797 * bug fix: division operators (MOD, TRUNCATE and the like) with
798 constant zero divisors and integer dividends no longer generate
800 * bug fix: provide default methods for INPUT-STREAM-P and
801 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
802 * bug fix: improve the detection and resolution of MOP metacycles
803 (where effective method computation depends on the generic function
804 having its effective method computed). (reported by Bruno Haible)
805 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
806 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
807 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
808 * bug fix: pathnames with relative directory components can be
809 represented relative to default pathnames.
810 * optimization: in taking the GCD of bignums, reduce the two bignums
811 to approximately the same size (using Euclid's algorithm) before
812 applying the more sophisticated binary GCD. (thanks to Juho
814 * optimization: COUNT on bitvectors now operates word-at-a-time.
815 * optimization: ASH with a positive, but not necessarily constant,
816 (leftwards) shift, when in a modular context, is compiled to a
818 * fixed some bugs revealed by Paul Dietz' test suite:
819 ** FORMAT variable parameters ("~V<char>") are defaulted properly
820 if the corresponding argument is NIL.
821 ** FORMAT directives accepting a minpad parameter treat negative
823 ** literal commas as character parameters to FORMAT directives are
825 ** literal spaces directly after ~<Newline> directives within a
826 format-logical-block (~:< ~@:>) do not induce :FILL-style
827 conditional newlines.
828 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
829 as specified: it no longer includes conditional newlines).
830 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
832 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
833 :PER-LINE-PREFIX argument does not evaluate to a string.
835 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
836 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
837 the manual for details; for now, package locks can be disabled by
838 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
839 if no major problems are found then it is likely that they will be
840 compiled in unconditionally.
841 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
842 unsupported operators on all platforms. To load a shared library
843 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
844 object file, link it into a shared library outside of SBCL and
845 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
847 * bug workaround: ROOM T and the GC were not working together
848 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
849 reporting facility aren't preserved. That reporting has been
850 disabled, so now until and unless someone figures out how to make it
851 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
852 * minor incompatible change: as threatened around sbcl-0.8.0, the
853 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
854 The symbols are also exported from SB-PCL for backwards
855 compatibility, but more so than before SB-PCL should be treated as
856 an implementation-internal package.
857 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
859 * the SB-SPROF contrib now works on (most) non-x86 architectures.
860 It is known as of this release not to work on the Alpha, however.
861 * fixed bug #167: errors signalled due to illegal syntax in method
862 bodies are now more legible.
863 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
864 designators and can hence be used with TYPEP.
865 * fixed bug #333: CHECK-TYPE now ensures that the type error
866 signalled, if any, has the right object to be accessed by
867 TYPE-ERROR-DATUM. (reported by Tony Martinez)
868 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
869 5.1.2.3 for multiple-value place subforms. (reported by Kalle
871 * fixed bug #334: programmatic addition of slots using specialized
872 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
873 and :ALLOCATION :CLASS slots.
874 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
875 float exponents. (rereported by Peter Seibel)
876 * fixed a bug: #\Space (and other whitespace characters) are no
877 longer considered to be macro characters in standard syntax by
879 * fixed bug: initialization of condition class metaobjects no longer
880 causes an instance of the condition to be created. (reported by
882 * fixed bug: it is now possible to have slots such that
883 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
884 description is neither :INSTANCE nor :CLASS.
885 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
886 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
887 SLOT-VALUE-USING-CLASS), and no longer causes errors with
888 non-standard SLOT-DEFINITION-ALLOCATION values.
889 * fixed bugs: various race conditions handling exiting threads.
890 CL-PPCRE's thread torture test now passes.
891 * fixed arguably-a-bug: GC time proportional to number of threads in
892 system even when most of them are idle
893 * optimization: improved performance of BIT and SBIT on bit-vectors.
894 * .fasl file incompatibility: The fasl file version number has
895 been incremented because of changes associated with package locks.
897 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
898 * minor incompatible change: the system no longer provides
899 optimization hints (conditions which are TYPEP
900 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
901 mode; these hints are still emitted when higher SPEED optimization
903 * new contrib module: a sampling profiler (profiling by statistical
904 sampling, rather than by instrumenting functions) is available as
905 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
906 * the behaviour of the standard function ED is now customizeable by
907 third parties through a hook variable: see ED's documentation
908 string for information on the protocol.
909 * the compiler no longer emits efficiency notes for (FUNCALL X)
910 when the type of X is uncertain under default optimization
912 * fixed bug 276: mutating a binding of a specialized parameter to a
913 method to something that is not TYPEP the specializer is now
915 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
916 exhibit the required behaviour on the x86 platform. (thanks to
917 Peter van Eynde, Eric Marsden and Bruno Haible)
918 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
919 even for difficult arguments. (reported by Peter Graves)
920 * fixed bug 141a: the backquote printer now descends quoted
922 * fixed another bug in backquote printing: no more destructive
923 modification of the form's list structure. (reported by Brian
925 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
926 move between its address being taken and the call to
927 interrupt_thread, fixing a crashing race condition.
928 * the SB-POSIX contrib implementation has been adjusted so that it
929 no longer exhibits ridiculously poor performance when constructing
930 instances corresponding to C structs.
932 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
933 * minor incompatible change: the sb-grovel contrib now treats C
934 structures as alien (in the sense of SB-ALIEN) objects rather than
935 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
936 has implications for memory management of client code
937 (sb-grovel-returned objects must now be manually managed) and for
938 type safety (alien objects now have full types).
939 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
940 used to control emission of compiler diagnostics, rather than the
941 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
942 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
943 quality should be considered deprecated.
944 * install.sh now installs the user manual as well
945 * (not quite a new documentable feature, but worth considering in
946 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
947 beginnings of a semantically meaningful condition hierarchy is
948 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
950 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
951 designator as the defaults argument.
952 * fixed bug: Displaced arrays whose displaced-to array has become
953 too small now cause ARRAY-DIMENSION to signal an error, providing
954 for safer bounds-checking. (reported by Bruno Haible)
955 * fixed bug: DEFCLASS slot definitions with identical :READER and
956 :WRITER names now signal a reasonable error. (reported by Thomas
958 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
960 * fixed bug: Invalid slot specification errors now print correctly.
961 (thanks to Zach Beane)
962 * fixed bug 320: Shared to local slot value transfers in class
963 redefinitions now happen corrently from superclasses as
964 well. (reported by Bruno Haible)
965 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
967 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
968 improper lists correctly. (reported by Bruno Haible)
969 * fixed bug 313: source-transform for <fun-name> was erroneously
970 applied to a call of a value of a variable with name <fun-name>.
971 (reported by Antonio Menezes Leitao)
972 * fixed bug 307: The obsolete instance protocol ensures that
973 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
974 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
975 remove dead unknown-values globs from the middle of the stack.
976 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
977 easier to resume long computations after using *BREAK-ON-SIGNALS*
978 to diagnose and fix failures. (thanks to Nikodemus Siivola)
979 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
980 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
981 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
982 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
983 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
985 * fixed a (fairly theoretical) bug in string printing: if
986 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
987 does not have array-element-type equal to the most general string
989 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
990 function name. (thanks to Nikodemus Siivola)
991 * fixed bug: compile-time format string checker failed on
992 non-closed ~{. (reported by Thomas F Burdick)
993 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
994 don't warn on legal loop code involving more than one aggregate
995 boolean. (thanks to Nikodemus Siivola)
996 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
997 in keyword position to MAKE-PACKAGE was not being reported as
999 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
1000 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
1002 * fixed bugs 280 and 312: the checking for multiple definitions in a
1003 file is less likely to become confused by uses of inline
1005 * fixed bug: the #S reader macro performs the keyword coercion
1006 specified for slot names. (reported by Kalle Niemitalo)
1007 * fixed bug: lambda lists may contain symbols whose names start with
1008 & but are not lambda-list-keywords; their occurrence triggers a
1010 * fixed bug 321: define-method-combination argument lambda lists do
1011 not cause invalid code to be generated when &optional or &aux
1012 variables have default values. (reported by Bruno Haible)
1013 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
1014 have CLOS classes; this appears also to have fixed reported
1015 instability in stack exhaustion detection.
1016 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
1017 MUFFLE-WARNING when no associated restart is present is now
1019 * optimization: rearranged the expansion of various defining macros
1020 so that each expands into only one top-level form in a
1021 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
1023 * optimization: used a previously little-used slot in symbols to
1024 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
1026 * fixed some bugs revealed by Paul Dietz' test suite:
1027 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
1029 ** Fixed an optimization bug related to inheritance of initforms
1030 from local to shared slots.
1031 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
1032 ** CAST optimizer forgot to flush argument derived type.
1033 ** print/read consistency on floats is now orders of magnitude
1034 more likely. (thanks also to Bruno Haible for a similar report
1036 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
1037 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
1038 for objects of type REAL. Make it so.
1039 ** SXHASH is specified (infelicitously) to respect similarity,
1040 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
1041 it so. (thanks to Markus Ziegler)
1042 ** on the Alpha, the compiler succeeds in compiling functions
1043 returning a known number of arguments greater than 63.
1044 ** fixed handling of invalid NIL arguments in keyword position
1046 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
1048 ** provided a readably-printable representation for RANDOM-STATE
1050 ** ensured that pathnames, where they have a namestring, always
1051 print using #P"..." syntax.
1053 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
1054 * Support for the forthcoming 2.0 version of the NetBSD kernel
1055 running on x86 hardware has been added. (thanks to Perry
1056 E. Metzger most immediately, and others for their past work)
1057 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
1058 current release version; SBCL's OpenBSD support had been broken
1059 since about the time of OpenBSD's switch to ELF binary format.)
1060 * [placeholder for DX summary]
1061 ** user code with &REST lists declared dynamic-extent, under high
1062 speed or space and low safety and debug optimization policy.
1063 * The manual has been converted to Texinfo format and the debugger
1064 chapter from the cmucl manual has been added.
1065 * A facility has been added to extract documentation strings from
1066 sbcl and store them as Texinfo-formatted snippets for inclusion in
1067 the manual (via Texinfo's @include directive)
1068 * bug fix: compiler emitted division in optimized DEREF. (thanks for
1069 the test case to Dave Roberts)
1070 * bug fix: multidimensional simple arrays loaded from FASLs had fill
1071 pointers. (reported by Sean Ross)
1072 * bug fix: PROFILE output is printed nicely even for large numerical
1073 values. (thanks to Zach Beane)
1074 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
1075 greater than 32 handle EOF correctly.
1076 * bug fix: on X86 an immediate argument of the IMUL instruction is
1077 correctly printed in disassembly. (thanks to Lutz Euler)
1078 * bug fix: class slots in redefined classes preserve their old
1079 values. (thanks to Bruno Haible and Nikodemus Siivola)
1080 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
1081 succeeds. (reported by Marco Baringer)
1082 * fixed some bugs revealed by Paul Dietz' test suite:
1083 ** READ-SEQUENCE now works on ECHO-STREAMs.
1084 ** RATIONALIZE works more according to its specification. (thanks
1087 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
1088 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
1089 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
1090 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
1091 should matter to you only if you rebind the printer control
1092 variables and then find you want different bindings in the
1093 debugger than in the ordinary execution of your program.)
1094 * The runtime build system has been tweaked to support building
1095 (on SPARC/SunOS) using a C compiler which invokes Sun's own
1096 assembler and linker. (thanks to Nikodemus Siivola)
1097 * Unbound, undefined, undeclared variables now trigger full
1098 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
1099 more often programmer error than deliberate exploitation of undefined
1101 * optimization: the hash algorithm for strings has changed to one
1102 that is less vulnerable to spurious collisions. (thanks to Juho
1104 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
1105 needless bounds checking. (thanks to Juho Snellman)
1106 * optimization: implemented multiplication as a modular
1107 (UNSIGNED-BYTE 32) operation on the PPC backend.
1108 * fixed some bugs revealed by Paul Dietz' test suite:
1109 ** ADJUST-ARRAY now copies the datum in a zero rank array if
1111 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
1112 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
1113 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
1115 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
1117 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
1119 ** ECHO-STREAMs no longer attempt to echo the end of file value to
1120 their output stream on EOF from read.
1121 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
1122 have been read to end-of-file.
1123 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
1125 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
1126 description of determination of which consecutive characters
1128 ** Printing the "Space" character with escaping on now yields "#\\ ",
1129 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
1130 ** Reading floating-point numbers with *READ-BASE* set to a number
1131 less than 10 works correctly.
1132 ** Reading floating-point numbers with *READ-BASE* set to a number
1133 more than 10 works correctly.
1134 ** Printing with *PRINT-READABLY* targets the standard readtable, not
1135 the readtable currently in effect.
1137 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
1138 * minor incompatible change: parsing of namestrings on a physical
1139 (Unix) host has changed; numbers after the final #\. in a
1140 namestring are no longer interpreted as a version field. This is
1141 intented to be largely invisible to the user, except that the
1142 meaning of the namestring "*.*.*" has changed: it now refers to a
1143 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
1144 should usually be replaced by
1145 (make-pathname :name :wild :type :wild :version :wild)
1146 with the added benefit that this is more likely to be portable.
1147 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
1148 option to OPEN now signals an error if the file being opened
1149 exists; this may have an impact on existing code.
1150 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
1151 signals received, on the PowerPC platforms (both Linux and
1152 Darwin). (thanks to Pierre Mai for pointing out the location of
1154 * several fixes on OS X: The system now builds and runs cleanly on
1155 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
1156 (thanks to Brian Mastenbrook)
1157 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
1158 second-resolution: integer multiples of 1/3600 between -24 and 24.
1159 (thanks to Vincent Arkesteijn)
1160 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
1161 type when called with 1 argument; PEEK-CHAR checked type of
1162 PEEK-TYPE only after having read first character from a
1163 stream. (reported by Peter Graves)
1164 * bug fix: the garbage collector now has much better locality
1165 behaviour, and in particular no longer treats all memory as being
1166 exhausted when many small objects point to each other in a deeply
1168 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
1169 recognized as being TYPEP their class.
1170 * bug fix: the PUSHNEW documentation string has been corrected.
1171 (thanks to Vincent Arkesteijn)
1172 * bug fix: defaulting of the value for the last of an atypically
1173 large number of multiple values being bound was not being
1174 performed correctly on the Alpha or PPC platforms
1175 * optimization: implemented multiplication as a modular
1176 (UNSIGNED-BYTE 32) operation on the x86 backend.
1177 * optimization: SEARCH on simple-base-strings can now be open-coded.
1178 (see also contrib/compiler-extras.lisp for inspiration for
1179 teaching the compiler about the Boyer-Moore algorithm).
1180 * value, returned by MAX (and MIN) called with several EQUALP, but
1181 not EQL, arguments now does not depend on compiler settings.
1182 * fixed some bugs revealed by Paul Dietz' test suite:
1183 ** in stack analysis liveness information is propagated from
1184 non-local entry points.
1185 ** pathwise CAST removing failed when the CAST node did not start
1187 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
1188 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
1190 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
1191 host is already defined.
1192 ** RENAME-FILE works on streams instead of signalling an internal
1194 ** PEEK-CHAR uses the current readtable when determining whether
1195 or not a character is whitespace.
1196 ** MERGE-PATHNAMES handles the case when the pathname does not
1197 specify a name while the default-pathname specifies a version.
1198 ** Pathnames now stand a better chance of respecting print/read
1200 ** Attempting to use standardized file system operators with a
1201 pathname with invalid :DIRECTORY components signals a
1203 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
1205 ** DIRECTORY on logical pathnames is more correct.
1206 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
1207 signal a TYPE-ERROR if their argument is not a stream.
1208 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
1209 designator argument does not designate a stream.
1210 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
1211 examining the synonym.
1212 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
1214 ** OPEN and WITH-OPEN-STREAM allow opening streams with
1215 element-type larger than ([UN]SIGNED-BYTE 32).
1217 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
1218 * When built with the :SB-FUTEX feature, threaded builds now take
1219 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
1220 for faster/more reliable mutex and condition variable support.
1221 * Incompatible change (but one you probably shouldn't have been using
1222 anyway): the interface and code for arbitrating between multiple
1223 threads in the same user session has been redesigned.
1224 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
1225 error when called without an explicit environment argument.
1226 (thanks to Nikodemus Siivola)
1227 * bug fix: buffered :DIRECTION :IO streams are less likely to become
1228 confused about their position. (thanks to Adam Warner and Gerd
1230 * bug fix: Pretty printing backquoted forms with unquotations in the
1231 argument list position of various code constructs such as LAMBDA
1232 now works correctly. (reported by Paul Dietz)
1233 * bug fix: Pretty printing unquotations no longer loses all
1234 stream position information.
1235 * optimization: performance of string output streams is now less
1236 poor for multiple small sequence writes.
1237 * optimization: performance of CSUBTYPEP in the presence of complex
1238 expressions involving CONS and NOT many times has been improved.
1239 (reported by Paul Dietz)
1240 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
1242 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
1243 streams for the REPL, for the debugger, and so on. (thanks to
1245 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
1246 * fixed some bugs revealed by Paul Dietz' test suite:
1247 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
1248 is ignored if it is too small, rather than propagating through
1249 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
1250 ** extremely complex negations of CONS types were not being
1251 sufficiently canonicalized, leading to inconsistencies in
1253 ** VALUES tranformer lost derived type.
1255 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
1256 * fixed a bootstrapping bug: the build process no longer assumes
1257 that the various BOOLE-related constants have the same value in
1258 host and target lisps. (noted by Paul Dietz' test suite on an
1259 SBCL binary built from CLISP)
1260 * The system can now be dynamically linked on the MIPS platform,
1261 which enables dynamic loading of foreign code from Lisp. (thanks
1262 to Ralf Baechle for discussions on the MIPS ABI)
1263 * The system now records debugging information for its own source
1264 files in a filesystem-position-independent manner, relative to
1265 the "SYS" logical host.
1266 * fixed a compiler bug: MV-LET convertion did not check references
1267 to the "max args" entry point. (reported by Brian Downing)
1268 * tweaked disassembly notes to be less confident about proclaiming
1269 some instruction as an LRA. (thanks to Brian Downing)
1270 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
1271 now each have their own history, command character, and other
1272 characteristics. (thanks to David Lichteblau)
1273 * fixed some bugs revealed by Paul Dietz' test suite:
1274 ** compiler failure in compiling LOGAND expressions including a
1276 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
1277 shift greater than 32.
1278 ** FLUSH-DEST did not mark blocks for type check regeneration.
1279 ** HANDLER-CASE failed to accept declarations in handler clauses
1280 in some circumstances.
1282 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
1283 * New code in contrib/sb-introspect (still probably not entirely
1284 stable yet) provides some support for smart Lisp development
1285 environments like SLIME.
1286 * The conditions signalled for errors occurring when loading .fasl
1287 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
1288 in a way which should help ASDF recover gracefully.
1289 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
1290 stuff has been cleaned up. If you code contrib/ stuff, this might
1291 affect you, and you can look at contrib/README, contrib/STANDARDS,
1292 and/or the 0.8.4.27 diff to check.
1293 * In full calls the compiler now does not generate checks for declared
1294 argument types for all arguments.
1295 * various threading fixes
1296 ** and some experimental patches which didn't make it into
1297 the main tree for this release, but which are shipped in
1298 contrib/experimental-thread.patch as a possible fix for some
1299 failures (deadlock, spinning...) in GC-intensive multithreaded
1301 * fixed PPC build problem (source code incompatibility of different
1302 library versions): added offsetof() hackery which attempts to divine
1303 where glibc maintainers put uc_mcontext today
1304 * fixed bug 282: compiler does not trust type assertions while passing
1305 arguments to a full call.
1306 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
1307 &OPTIONAL VALUES type parameter.
1308 * fix bug 214: algorithm for noting rejected templates is now more
1309 similar to that of template seletion. (also reported by rydis on
1311 * fixed bug 141b: printing backquoted information readably and prettily
1312 inserts a space where necessary.
1313 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
1314 (CHARACTER 10) are now reported as errors, rather than propagated
1315 as unknown types. (reported by piso on #lisp)
1316 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
1317 with values NIL and :ERROR. (thanks to Milan Zamazal)
1318 * fixed bug 191c: CLOS now does proper keyword argument checking as
1319 described in CLHS 7.6.5 and 7.6.5.1.
1320 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
1321 longer raise an error; further, using a list as a for-as-arithmetic
1322 counter now raises a meaningful error.
1323 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
1324 now understood by sequence creation functions such as MAKE-SEQUENCE
1326 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
1327 a STRING-INPUT-STREAM.
1328 * compiler enhancement: SIGNUM is now better able to derive the type
1330 * type declarations inside WITH-SLOTS are checked. (reported by
1332 * fixed some bugs revealed by Paul Dietz' test suite:
1333 ** incorrect optimization of TRUNCATE for a positive first
1334 argument and negative second.
1335 ** compiler failure in let-convertion during flushing dead code.
1336 ** compiler failure while deriving type of TRUNCATE on an
1337 interval, containing 0.
1338 ** ASH of a negative bignum by a negative bignum count now returns
1340 ** intersection of CONS types now canonicalizes properly, fixing
1341 inconsistencies in SUBTYPEP.
1343 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
1344 * incompatible change: The --disable-debugger command line
1345 option now clobbers the debugger at a more fundamental
1346 level, by redefining #'INVOKE-DEBUGGER instead of by
1347 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
1348 is specified by ANSI to ignore *DEBUGGER-HOOK* and
1349 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
1350 enter the debugger REPL and then suffer recursive errors
1351 because *DEBUG-IO* is also messed up in --disable-debugger mode;
1352 while under the new system, BREAK in --disable-debugger mode
1353 terminates the system just as an unhandled error would.
1354 * fixed compiler performance when processing loops with a step >1;
1355 * bug fix: DOCUMENTATION now retrieves generic function
1356 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
1357 support has been systematized, and now supports the methods
1358 specified by ANSI, along with a default method and a method for
1359 slot documentation. (reported by Nathan Froyd)
1360 * bug fix: effective methods associated with a generic function are
1361 no longer cached over a change of that generic function's method
1362 combination. (reported by Andreas Fuchs)
1363 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
1364 names a non-existent directory. (thanks to Andreas Fuchs)
1365 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
1366 the PPC platform, lead to stack corruption; this has been fixed.
1367 (reported by Rainer Joswig)
1368 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
1369 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
1370 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
1371 arithmetic in general, is now much more reliable on the PPC
1373 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
1374 platform now returns the right answer.
1375 * optimization: restored some effective method precomputation in
1376 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
1377 precomputation is now tunable.
1378 * optimization: compiler-internal data structure use has been
1379 reviewed, and changes have been made that should improve the
1380 performance of the compiler by about 20%.
1381 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
1382 simple-base-strings and simple-bit-vectors is improved.
1383 * optimization: the optimization of 32-bit logical and arithmetic
1384 functions introduced in version 0.8.3 on the x86 has been
1385 implemented on the mips, ppc and sparc platforms; an
1386 implementation of the same facility, but for 64-bit arithmetic,
1387 has been added for the alpha.
1388 * microoptimization: the compiler is better able to make use of the
1389 x86 LEA instruction for multiplication by constants.
1390 * bug fix: in some situations compiler did not report usage of
1391 generic arithmetic in (SPEED 3) policy.
1392 * bug 145b fix: compiler used wrong type specifier while converting
1393 MEMBER-types to numeric.
1394 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
1396 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
1397 index argument now works properly on the Alpha platform.
1398 * bug fix: floating point exception treatment on the Alpha platform
1400 * bug fix: FILE-POSITION works much better on string input and
1401 output streams. (thanks to Nikodemus Siivola)
1402 * bug fix: many threading/garbage collection symptoms sorted.
1403 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
1404 might be pseudo-atomic.
1405 * internal change: Stopping for GC is now done with signals not
1406 ptrace. GC is now done in whichever thread wanted it, instead of
1408 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
1410 * bug fix: result form in DO is not contained in the implicit
1412 * incompatible change: ICR structure is changed; the value part of
1413 CONTINUATION is now called LVAR; corresponding functions are
1414 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
1415 * added type deriver for ISQRT (thanks to Robert E. Brown).
1416 * bug fix: better support for loading from the command line when an
1417 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
1419 * fixed some bugs revealed by Paul Dietz' test suite:
1420 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
1421 ** ROUND and FROUND now give the right answer when given very
1422 small float arguments.
1423 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
1425 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
1426 ** (GCD 0 <negative-integer>) returned <negative-integer>.
1427 ** LCM should return a non-negative integer.
1428 ** PARSE-INTEGER returned the index of a terminator instead of the
1429 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
1430 ** PARSE-INTEGER returned an incorrect index being applied to a
1432 ** LCM with two arguments of 0 returns 0 rather than signalling
1434 ** unsigned addition of a 32-bit constant with the high bit set no
1435 longer causes an internal compiler error on the x86.
1436 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
1437 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
1438 with negative last argument.
1439 ** byte specifiers with zero size and position no longer cause
1440 an error during type derivation.
1441 ** bignum multiplication on the Alpha platform now returns the
1443 * porting: The system now builds on SuSE AMD64, although it still
1444 generates a 32-bit binary.
1445 * .fasl file incompatibility: The fasl file version number has
1446 been incremented (because of the changes to internal compiler
1447 data structures referred to above).
1449 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
1450 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
1451 more accurately, on the Darwin kernel running on PowerPC hardware.
1452 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
1453 * Compiler code deletion notes now signal a condition of type
1454 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
1455 an associated MUFFLE-WARNING restart.
1456 * The compiler now performs limited argument count validation of
1457 constant format strings in FORMAT, and where appropriate in ERROR,
1458 CERROR and WARN. (thanks to Gerd Moellmann)
1459 * New ASDF-INSTALL contrib can be used for automatic download and
1460 installation of third-party Lisp code from CCLAN or other sites
1462 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
1463 which forces another thread to execute a function supplied by the
1465 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
1466 now accept and act upon their :ELEMENT-TYPE keyword argument.
1467 (reported by Martin Atzmueller, Edi Weitz)
1468 * bug fix: FILE-POSITION now accepts position designators up to
1469 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
1470 is the greater. (thanks to Patrik Nordebo)
1471 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
1472 Kalle Olavi Niemitalo)
1473 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
1474 printable. (reported by Eric Marsden)
1475 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
1476 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
1477 * x86 bug fix in control stack exhaustion checking: now shows backtrace
1478 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
1479 (thanks to Stig Sandoe)
1480 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
1481 UNREAD-CHAR now correctly includes the unread character in the
1482 target sequence. (thanks to Gerd Moellmann)
1483 * bug fix in threaded builds: the system can now be suspended and
1484 resumed by shell job control with minimal disruption.
1485 * bug fixes in times and timezones >2038AD
1486 * better handling of "where is GNU make?" problem in build scripts
1487 (thanks to Nikodemus Siivola)
1488 * new optimization: inside a named function any reference to a
1489 function with the same name is considered to be a self-reference;
1490 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
1491 optimization quality.
1492 * new optimization on x86: logical functions and + now have
1493 optimized (UNSIGNED-BYTE 32) versions, which are automatically
1494 used when the result is truncated to 32 bits.
1495 * VALUES declaration is partially enabled.
1496 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
1497 * fixed some bugs revealed by Paul Dietz' test suite:
1498 ** The system now obeys the constraint imposed by
1499 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
1500 types form a lattice under type intersection.
1501 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
1502 ** ASSOC now ignores NIL elements in an alist.
1503 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
1504 and (1+ MOST-POSITIVE-FIXNUM) answers.
1505 ** The addition of a method with invalid qualifiers to a generic
1506 function does not cause an error to be signalled immediately;
1507 a warning is signalled, and the error is generated only on
1508 calling the generic function.
1509 * changed .fasl file version number, in part to add type codes for
1510 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
1511 obscure ANSI requirements
1513 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
1514 * fixed bug 148: failure to inline-expand a local function left
1515 garbage, confusing the compiler.
1516 * fixed bugs 3cd: structure slot readers perform type check if the
1517 slot can have an invalid value (i.e. it is either not initialized
1518 or can be written with a less specific slot writer).
1519 * bug fix: the compiler now traps array references to elements off
1520 the end of an array; previously, the bounds checking in some
1521 circumstances could go off-by-one.
1522 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
1524 * type declarations for array element types now obey the description
1525 on the CLHS page "Declaration TYPE", as per discussions on
1526 sbcl-help around 2003-05-08. This means that a declaration
1527 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
1528 declaration, all references to BAR will be asserted or assumed
1529 (with THE, so dependent on compiler policy) to involve objects of
1530 type FOO. Note that no such declaration is implied in
1531 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
1532 * declared types of functions from the "Conditions"
1533 chapter. (reported by Paul Dietz)
1534 * bug fix: CERROR accepts a function as its first argument.
1535 * bug fix: NTH an NTHCDR accept a bignum as index
1536 arguments. (reported by Adam Warner)
1537 * optimization: character compare routines now optimize comparing
1538 against a constant character. (reported by Gilbert Baumann)
1539 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
1540 argument now works properly on the MIPS platform.
1541 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
1542 * repeated evaluation of the same DEFSTRUCT, a slot of which is
1543 declared to have a functional type, does not cause an error
1545 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
1546 many :DEFAULT-INITARGS. (reported by Istvan Marko)
1547 * fixed bug: if last continuation of a deleted block has a
1548 destination, this destination should be deleted too. (reported by
1550 * fixed a bug in the bootstrap process: the host compiler's values
1551 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
1552 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
1553 test case from Patrik Nordebo)
1554 * improved the ability of the disassembler on the PPC platform to
1555 provide helpful disassembly notes.
1556 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
1557 the class in more cases than previously.
1558 * bug fix: FILE-POSITION now understands :START and :END for
1559 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
1560 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
1561 8) data. (thanks to David Lichteblau)
1562 * bug fix: it is possible to add a method to a generic function
1563 without lambda list.
1564 * bug fix: reader failed to signal END-OF-FILE inside an
1565 object representation. (reported by Nikodemus Siivola)
1566 * fixed some bugs revealed by Paul Dietz' test suite:
1567 ** LAST and [N]BUTLAST should accept a bignum.
1568 ** condition slot accessors are methods.
1569 ** (VECTOR NIL) is a subtype of STRING.
1571 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
1572 * minor incompatible change: some nonsensical specialized lambda
1573 lists (used in DEFMETHOD) which were previously ignored now signal
1575 * minor incompatible change: the system is now aware of the types of
1576 variables in the COMMON-LISP package, and will signal errors for
1577 most violations of these type constraints (where previously they
1578 were silently accepted).
1579 * minor incompatible change: COMPILE-FILE now uses the freedom
1580 afforded (ANSI 3.2.2.3) to use derived function types for
1581 functions defined in the same file. This also permits the system
1582 to warn on static type mismatches and function
1583 redefinition. (Currently it does not work with high DEBUG level.)
1584 * minor incompatible change: VALUES declaration is disabled.
1585 * When issuing notes, the compiler now signals a condition of type
1586 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
1587 restart for use in user handlers. It is expected that the
1588 COMPILER-NOTE condition will eventually become a condition
1589 supertype to a hierarchy of note types, which will then be
1590 handleable in a similar fashion. However, at the moment, no such
1591 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
1592 but it's an implementation detail, not a classification for the
1594 * Changes in type checking closed the following bugs:
1595 ** type checking of unused values (192b, 194d, 203);
1596 ** template selection based on unsafe type assertions (192c, 236);
1597 ** type checking in branches (194bc).
1598 * A short form of VALUES type specifier has ANSI meaning (it has
1599 increased the number of situations when SBCL cannot perform type
1601 * fixed bug in DEFSTRUCT: once again, naming structure slots with
1602 keywords or constants is permissible.
1603 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
1604 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
1605 classes. (thanks to Antonio Martinez)
1606 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
1607 outputs FRESH-LINE or TERPRI, and no longer converts its stream
1608 argument to a pretty-print stream. Instead, it leaves any such
1609 operations to DESCRIBE-OBJECT methods.
1610 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
1612 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
1613 lambda lists. (thanks to David Lichteblau)
1614 * bug fix: Defining a generic function with a :METHOD-CLASS being a
1615 subclass of STANDARD-METHOD no longer causes stack exhaustion.
1616 (thanks to Gerd Moellmann)
1617 * fixed bug 246: increased compilation speed of long
1618 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
1620 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
1622 * DEFCONSTANT now throws a condition of type
1623 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
1624 constant to a non-EQL value; CONTINUE and ABORT restarts
1625 respectively change and preserve the value.
1626 * fixed bug 63: The code walker, part of the implementation of CLOS,
1627 is now better at handling symbol macros.
1628 * bug fix: There is no longer an internal implementation type named
1629 CL:LENGTH. (reported by Raymond Toy)
1630 * bug fix: In macro-like defining macros/special operators the
1631 implicit block does not enclose the lambda list.
1632 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
1633 enclosed in parentheses) are not suitable as type specifiers, and
1634 their use properly signals an error now.
1635 * bug fix: An argument count mismatch for a type specifier in code
1636 being compiled no longer causes an unhandled error at compile
1637 time, but signals a compile-time warning.
1638 * fixed simple vector readable printing
1639 * bug fix: DESCRIBE takes more care over whether the class
1640 precedence list slot of a class is bound before accessing it.
1641 (reported by Markus Krummenacker)
1642 * bug fix: FORMATTER can successfully compile pretty-printer format
1643 strings which use variants of the ~* directive inside.
1644 * bug fix: SEARCH now applies its TEST predicate to the elements of
1645 the arguments in the correct order. (thanks to Wolfhard Buss)
1646 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
1647 in type inference. (thanks to Robert E. Brown)
1648 * bug fix: Reading in symbols with an explicit package name of ""
1649 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
1650 * fixed some bugs revealed by Paul Dietz' test suite:
1651 ** NIL is now allowed as a structure slot name.
1652 ** Arbitrary numbers, not just REALs, are allowed in certain
1653 circumstances in LOOP for-as-arithmetic clauses.
1654 ** Multiple class redefinitions before slot access no longer
1655 causes a type error.
1656 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
1657 association between the name and a class.
1658 ** Generic functions with non-standard method-combination and over
1659 five methods all of which return constants no longer return NIL
1660 after the first few invocations. (thanks to Gerd Moellmann)
1661 ** CALL-NEXT-METHOD with no arguments now passes the original
1662 values of the arguments, even in the presence of assignment.
1663 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
1665 ** DISASSEMBLE works with closures and funcallable instances.
1666 ** ADD-METHOD now returns the generic function, not the new method.
1667 ** FIND-METHOD signals an error if the lengths of the specializers
1668 is incompatible with the generic function, even if the ERRORP
1670 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
1671 which its argument is a member.
1672 ** DEFCLASS only redefines the class named by its class-name
1673 argument if that name is the proper name of the class;
1674 otherwise, it creates a new class.
1675 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
1676 of the UNBOUND-SLOT condition to the name of the slot.
1677 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
1678 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
1679 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
1680 SLOT-MAKUNBOUND in the specified fashion.
1682 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
1683 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
1684 cross-compilation host. As a consequence, we can now bootstrap our
1685 way up to SBCL starting with a bare gcc toolchain and human-readable
1686 source code (first the source to CLISP, then the source to SBCL).
1687 * A contributed module containing a partial implementation of the
1688 simple-streams interface has been included. (thanks to Rudi
1690 * A contributed module implementing the RFC1321 Message Digest
1691 Algorithm, known as MD5, has been included.
1692 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
1693 no longer has any effect, as the code controlled by this feature
1694 has been deleted. (As far as we know, no-one has ever built using
1695 this feature, and its semantics were confused in any case).
1696 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
1697 work on conditions (as required by the ANSI specification),
1698 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
1699 expected behaviour on conditions. Users should note, however,
1700 that such behaviour is not required by the ANSI specification,
1701 and so use of this behaviour may render their code unportable.
1702 * fixed some bugs revealed by Paul Dietz' test suite:
1703 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
1705 ** &ENVIRONMENT parameter in macro lambda list is bound first.
1706 ** SXHASH on condition objects no longer returns NIL.
1707 ** :ALLOCATION :CLASS slots are better treated; their values are
1708 updated on class redefinition, and initforms inherited from
1709 superclasses are applied.
1710 ** REMOVE-METHOD returns its generic function argument even when
1711 no method was removed.
1712 ** SHARED-INITIALIZE now initializes the values of the requested
1713 slots, including those with :ALLOCATION :CLASS.
1714 ** ALLOCATE-INSTANCE now works on structure classes defined via
1715 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
1717 ** SLOT-EXISTS-P now works on conditions, as well as structures
1719 ** MAKE-LOAD-FORM now has the required methods on
1720 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
1721 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
1722 keyword, and now implements the SLOT-NAMES argument.
1723 ** methods with &OPTIONAL arguments no longer allow too many
1724 arguments to be passed in the call without error.
1725 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
1726 option is consistent with the required arguments of the generic
1727 function lambda list.
1728 * bug fix: REQUIRE accepts a string designator. (Thanks to
1730 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
1731 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
1732 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
1733 * bug fix: The compiler now checks for duplicated variables in macro
1735 * bug fix: SETQ on globals returns the correct value.
1736 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
1737 not a COMPILER-ERROR (followed by some other strange error on
1738 choosing the CONTINUE restart).
1739 * bug fix: make.sh and friends are now more consistent in the way that
1740 they look for GNU "make".
1742 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
1743 * experimental native threads support (on x86 Linux >=2.4 only).
1744 This is not compiled in by default: you need to add :SB-THREAD to
1745 the target features. See the "Beyond ANSI" chapter of the manual
1747 * fix for longstanding nonANSIism: The old distinction between
1748 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
1749 The return value from CL:FIND-CLASS is now a CLOS class, and
1750 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
1751 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
1752 * An interface to the MetaObject Protocol, as described in Kiczales,
1753 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
1754 MIT Press, 1991, is available from the SB-MOP package.
1755 * incompatible change: the SB-PCL package should now be considered
1756 a private implementation detail, and no longer a semi-private MOP
1758 * minor incompatible change: due to rearrangement for threads, the
1759 control stack and binding stack are now allocated at arbitrary
1760 addresses instead of being hardcoded per-port. Users affected by
1761 this probably have to be doing advanced things with shared
1762 libraries, and will know who they are.
1763 * minor incompatible change: Previously, all --eval forms used were
1764 processed with READ before any of them were processed with EVAL.
1765 Now each --eval form is processed with both READ and EVAL before
1766 the next --eval form is processed. (Thus package operations like
1767 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
1768 work as the user might reasonably expect.)
1769 * minor incompatible change: *STANDARD-INPUT* is now only an
1770 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
1772 * minor incompatible change: Y-OR-N-P is now character-oriented,
1773 not line oriented. Also, YES-OR-NO-P now works without errors.
1774 (thanks to Antonio Martinez)
1775 * sb-aclrepl module improvements: an integrated inspector, added
1776 repl features, and a bug fix to :trace command.
1777 * Known functions, which cannot be open coded by the backend, are
1778 considered to be able to check types of their arguments. (fixing
1779 a bug report by Nathan J. Froyd)
1780 * fixed a bug in computing method discriminating functions: It is
1781 now possible to define methods specialized on classes which have
1782 forward-referenced superclasses. (thanks to Gerd Moellmann)
1783 * fixed evaluation order in optional entries (reported by Gilbert
1785 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
1786 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
1787 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
1788 required-by-AMOP NAME argument, as well as CLASS and
1789 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
1790 * fixed bug 20: DEFMETHOD can define methods using names that are
1791 not the proper names of classes to designate class specializers.
1792 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
1793 Unix file descriptors, instead of blowing up. (thanks to Antonio
1795 * Garbage collection refactoring: user-visible change is that a
1796 call to the GC function during WITHOUT-GCING will not do garbage
1797 collection until the end of the WITHOUT-GCING. If you were doing
1798 this you were probably losing anyway.
1799 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
1800 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
1801 TYPEP the latter but not the former.
1802 * The compiler issues a full WARNING for calls to undefined functions
1803 with names from the CL package.
1804 * MAP-INTO for a vector destination is open coded. (reported by
1805 Brian Downing on c.l.l)
1806 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
1807 documentation string.
1808 * fixed some bugs revealed by Paul Dietz' test suite:
1809 ** COPY-ALIST now signals an error if its argument is a dotted
1811 ** Condition slots are now accessed more correctly in the presence
1812 of multiple initargs for a given slot.
1813 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
1814 exclude restarts of the same name associated with a different
1816 ** DEFCLASS of forward-referenced classes with another
1817 forward-referenced class in the superclasses list no longer
1819 ** Condition slots are now initialized once each, not multiple
1820 times. (thanks to Gerd Moellmann)
1821 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
1823 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
1824 ** NSET-EXCLUSIVE-OR does not return extra elements when its
1825 arguments contain duplicated elements.
1826 ** RESTART-CASE understands local macros.
1827 ** RESTART-CASE associates exactly its own restarts with a condition.
1828 ** ENDP in safe mode checks its argument to be of type LIST.
1829 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
1830 in question is unbound.
1831 ** Optimized MAKE-INSTANCE functions no longer cause internal
1832 assertion failures in the presence of duplicate initargs.
1833 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
1834 ** Side-effectful :DEFAULT-INITARGS have their side-effects
1835 propagated even in the ctor optimized implementation of
1837 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
1839 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
1840 * a better implementation of SXHASH on (simple) bit vectors,
1841 measured both in execution speed and in distribution of results
1842 over the positive fixnums, has been installed. Likewise, a better
1843 implementation of EQUAL for simple bit vectors is now available.
1844 * fixed CEILING optimization for a divisor of form 2^k.
1845 * fixed bug 240 (emitting extra style warnings "using the lexical
1846 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
1847 by Antonio Martinez)
1848 * fixed SXHASH, giving different results for NIL depending on type
1849 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
1850 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
1851 variable. (found by Rolf Wester)
1852 * SBCL does not ignore type declarations for special
1853 variables. (reported by rif on c.l.l 2003-03-05)
1854 * some bug fixes in contrib/sb-aclrepl/
1855 * fixed some bugs revealed by Paul Dietz' test suite:
1856 ** a bug in the CONS type specifier, whereby the CAR and CDR
1857 types got intertwined, has been fixed;
1858 ** the type system is now able to reason about the interaction
1859 between INTEGER and RATIO types more completely;
1860 ** APPEND, [N]REVERSE and NRECONC check that those their
1861 arguments, which must be proper lists, are really so;
1862 ** An array specialized to be unable to hold elements has been
1863 implemented, as required -- yes, really -- by ANSI;
1864 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
1865 on malformed property lists;
1867 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
1868 * incompatible packaging change: in line with Unix convention,
1869 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
1870 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
1871 * REQUIRE and PROVIDE are now optionally capable of doing something
1872 useful. See the documentation string for REQUIRE.
1873 * infrastructure for a managed SBCL contrib system: contributed
1874 modules in this release include:
1875 ** the ASDF system definition facility;
1876 ** an interface to the BSD Sockets API;
1877 ** an ACL-like convenience interface to the repl;
1878 (thanks to Kevin Rosenberg)
1879 ** an implementation of ROTATE-BYTE, with efficient implementation
1881 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
1882 gives a closer approximation to the right answer than previously.
1883 (thanks to Raymond Toy)
1884 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
1885 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
1886 optional environment arguments, as required by ANSI.
1887 * fixed bugs in other functions taking environment objects, allowing
1888 calls with an explicit NIL environment argument to be compiled
1890 * fixed bug 228: primary return values from
1891 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
1892 COMPILE or FUNCTION.
1893 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
1894 :TYPE will no longer signal errors on innocuous objects.
1895 * fixed bug 231b: SETQ is better at respecting type declarations in
1896 the lexical environment.
1897 * fixed a bug in DEFCLASS: classes named by symbols with no or
1898 unprintable packages can now be defined.
1899 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
1900 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
1901 * fixed compiler failure related to checking types of functions.
1902 (reported by Robert E. Brown)
1903 * the compiler is now much more consistent in its error-checking
1904 treatment of bounding index arguments to sequence functions: in
1905 (SAFETY 3) code, errors will be signalled in almost all cases if
1906 invalid sequence bounding indices are passed to functions defined
1907 by ANSI to operate on sequences.
1908 * fixed a bug in the build procedure: documentation of SBCL-specific
1909 packages is now preserved and available in the final Lisp image.
1910 * lifted FDEFINITION lookup out of loops in the implementation of
1911 many list operations. (thanks to Robert E. Brown)
1912 * fixed a bug in the reader: the #n# reader macro now works for
1913 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
1914 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
1915 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
1916 E. Brown and Tony Martinez respectively)
1917 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
1918 SLOT-BOUNDP: in particular, it is now able to optimize them much
1919 better, and is now not vulnerable to having packages renamed.
1920 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
1921 always called when a slot is not present in an instance. (thanks
1923 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
1924 longer signal an error on unreading a character following EOF on
1925 the previous constituent stream. (thanks to Tony Martinez)
1926 * fixed some bugs revealed by Paul Dietz' test suite:
1927 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
1928 not just nonnegative fixnums;
1929 ** the logical bit-array operators such as BIT-AND now accept an
1930 explicit NIL for their "opt-arg" argument (to indicate a
1931 freshly-consed result bit-array);
1932 ** ELT now signals an error on an invalid sequence index in safe
1934 ** the type system is now cleverer about negations of numeric
1935 types, and consequently understands the BIGNUM and RATIO types
1937 ** the type system is now cleverer about the interaction between
1938 INTEGER and RATIO types: while bugs still remain, many more
1939 cases are accurately computed;
1940 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
1941 if it is in the last clause;
1942 ** CONSTANTLY now correctly returns a side-effect-free function in
1944 ** DECLARE is no longer treated as a special-operator; in
1945 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
1946 * incremented fasl file version number due to the change in the
1947 DEFSTRUCT-SLOT-DESCRIPTION structure.
1949 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
1950 * minor incompatible change: code processed by the "interpreter" or
1951 EVAL now has a compilation optimization policy of (DEBUG 2)
1952 (changed from (DEBUG 1)) to improve debuggability of interactive
1953 development, and to allow the use of the debug RETURN command in
1955 * an experimental implementation of the RETURN command for the
1956 debugger has been included. (thanks to Frederik Kuivinen)
1957 * fixed bug 62: constraints were not propagated into a loop.
1958 * fixed bug in embedded calls of SORT (reported and investigated by
1960 * fixed some bugs revealed by Paul Dietz' test suite:
1961 ** printing and reading of arrays with some dimensions having
1962 length 0 (thanks to Gerd Moellmann);
1963 ** BOA constructor with &AUX argument without a default value does
1964 not cause a type error;
1965 ** CONSTANTP now returns true for all self-evaluating objects.
1967 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
1968 * fixed bug 127: DEFSTRUCT now does not clobber old structure
1969 accessors that are related by inheritance, as specified in the
1970 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
1971 Valtteri Vuorikoski)
1972 * The compiler is now able to inline functions that were defined in
1973 a complex lexical environment (e.g. inside a MACROLET).
1974 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
1976 * Support for the upcoming FreeBSD-5.0 release has been included.
1977 (thanks to Dag-Erling Smorgrav)
1978 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
1979 effect when it is not in a toplevel context.
1980 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
1981 stand a better chance of being correct. (thanks to Gerd
1983 * fixed bug in COERCE, which now signals an error on coercing a
1984 rational to a bounded real type which excludes the expected
1986 * The compiler is now able to derive types more accurately from the
1987 COERCE and COMPILE functions.
1988 * fixed bug 223: functional binding is considered to be constant
1989 only for symbols in the CL package.
1990 * fixed bug 231: SETQ did not check the type of a variable being set
1991 (reported by Robert E. Brown)
1992 * A new optimization for MAKE-INSTANCE has been included, fixing
1993 various bugs (including relating to :ALLOCATION :CLASS slots and
1994 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
1996 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
1997 ** As required by ANSI, LOOP now disallows anonymous collection
1998 clauses such as COLLECT I in conjunction with aggregate boolean
1999 clauses such as THEREIS (= I 1);
2000 ** LOOP now signals an error when any variable is reused in the
2001 same loop (including the potentially useful construct analogous
2002 to WITH A = 1 WITH A = (1+ A);
2003 ** IT is only a special loop symbol within the first clause of a
2004 conditional loop clause;
2005 ** LOOP with a typed iteration variable over a hashtable now
2006 signals a type error iff it should.
2007 * fixed some other bugs revealed by Paul Dietz' test suite:
2008 ** FILE-STREAM now names the class previously known as FD-STREAM;
2009 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
2010 argument) no longer signals an error;
2011 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
2012 of the slot symbol, rather than using the current package
2013 ((:CONC-NAME "") continues to intern the slot's name in the
2015 * incremented fasl file version number, because of the incompatible
2016 change to the DEFSTRUCT-DESCRIPTION structure, and again because
2017 of the new implementation of DEFINE-COMPILER-MACRO.
2019 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
2020 * Support for building SBCL for MIPS platforms running in
2021 little-endian mode has now been checked in, and basic
2022 functionality on said platforms verified.
2023 * minor incompatible change: PCL now records the pathname of a file
2024 in which methods and the like are defined, rather than its
2026 * minor incompatible change: TRUENAME now considers the truename of
2027 a file naming a directory to be the pathname with :DIRECTORY
2028 component indicating that directory.
2029 * minor incompatible change: a NAMED clause in the extended form of
2030 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
2031 reason for the previous behaviour is unclear.
2032 * more systematization and improvement of CLOS and MOP conformance
2033 in PCL (thanks to Gerd Moellman and Pierre Mai):
2034 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
2036 ** DEFINE-METHOD-COMBINATION no longer signals an error for
2037 primary methods with no specializers;
2038 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
2040 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
2041 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
2042 FORWARD-REFERENCED-CLASSes; error reporting on
2043 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
2045 ** SXHASH on CLOS instances now uses a slot internal to the
2046 instance to return different numbers on distinct instances,
2047 while preserving the same return value through invocations of
2049 ** DEFMETHOD signals errors when methods with longer incongruent
2050 lambda lists are added to generic functions;
2051 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
2052 CLASS, as specified in AMOP;
2053 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
2054 on the order returned by the primary method for classes of
2055 class STANDARD-CLASS;
2056 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
2057 * fixed some bugs shown by Paul Dietz' test suite:
2058 ** DOLIST puts its body in TAGBODY;
2059 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
2061 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
2062 value producing form;
2063 ** if more variables are given to PROGV than values, extra
2064 variables are bound and made to have no value;
2065 ** NSUBSTITUTE on list arguments gets the right answer with
2067 ** ELT signals an error of type TYPE-ERROR when the index argument
2068 is not a valid sequence index;
2069 ** LOOP signals (at macroexpansion time) an error of type
2070 PROGRAM-ERROR when duplicate variable names are found;
2071 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
2072 ** FUNCALL of special-operators now cause an error of type
2074 ** PSETQ now works as required in the presence of side-effecting
2075 symbol-macro places;
2076 ** NCONC accepts any object as its last argument;
2077 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
2079 ** loop-for-as-package does not require a package to be explicitely
2081 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
2083 * fixed bug 166: compiler preserves "there is a way to go"
2084 invariant when deleting code.
2085 * fixed bug 172: macro lambda lists with required arguments after
2086 &REST arguments now cause an error to be signalled. (thanks to
2088 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
2089 bug. (thanks to Matthew Danish)
2090 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
2092 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
2093 arguments are lexically rebound. (thanks to Gerd Moellmann and
2095 * fixed bug 194: error messages are now more informative when there
2096 is no primary method applicable in a call to a generic
2097 function. (thanks to Gerd Moellmann)
2098 * fixed bug in command line argument checking (thanks to Julian
2100 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
2101 (thanks to Matthew Danish)
2102 * incremented fasl file version number, because of the
2103 SXHASH-related changes in the layout of CLOS data structures
2105 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
2106 * minor incompatible change: The runtime (the Unix executable named
2107 "sbcl") is now much pickier about the .core files it will load.
2108 Essentially it now requires .core files to descend from the same
2109 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
2110 as the runtime does. (The intent is to prevent the crashes which
2111 can occur, and which can even be reported as mysterious failures,
2112 when people patch the sources or change the build parameters
2113 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
2114 sbcl and .core files.)
2115 * fixed bug: VALUES-LIST is no longer optimized away.
2116 * fixed bug 142: The FFI conversion of C string values to Lisp
2117 string values no longer conses excessively. (thanks to Nathan
2118 Froyd porting Raymond Toy's fix to CMU CL)
2119 * began to systematize and improve MOP conformance in PCL (thanks to
2120 Nathan Froyd, Gerd Moellman and Pierre Mai):
2121 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
2123 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
2124 ** FINALIZE-INHERITANCE is now called on class finalization;
2125 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
2126 argument precedence order.
2127 * fixed bug 202: The compiler no longer fails on functions whose
2128 derived types contradict their declared type.
2129 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
2130 so it can be non-toplevel.
2131 * The fasl file version number has changed (because of the new
2132 implementation of DEFMACRO).
2133 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
2134 safe code, that any length requirement by their type-specifier
2135 argument is valid. The exceptions to this are described in bug
2137 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
2138 functions now check the types of their inputs as required by ANSI.
2139 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
2140 introduced symbol is DECLAREd to be SPECIAL.
2141 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
2142 * fixed bug: :COUNT argument to sequence functions may be negative
2143 * fixed bug: body of DO-SYMBOLS may contain declarations
2144 * fixed bug: PUSHNEW now evaluates its arguments from left to right
2145 (reported by Paul F. Dietz, fixed by Gerd Moellman)
2146 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
2147 symbol macro only once
2148 * fixed printing of call frame when argument list is unavailable
2149 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
2150 * compiler no longer signals WARNING on unknown keyword
2153 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
2154 * A beta-quality port to the mips architecture running Linux,
2155 based on the old CMUCL backend, has been made. It has been tested
2156 on a big-endian kernel, and works sufficiently well to be able to
2157 rebuild itself; it has not been tested in little-endian mode.
2158 * fixed an inconsistency between gencgc.c and purify.c which made
2159 dumping/loading .core files unreliable
2160 * fixed bug 120a: The compiler now deals correctly with IFs where
2161 the consequent is the same as the alternative, instead of
2162 misderiving the return type. (thanks to Alexey Dejneka)
2163 * fixed bug 113: Logical pathnames are now dumpable (the logical
2164 host is resolved at load-time, throwing an error if it is not
2166 * fixed bug 174: FORMAT's error message is slightly clearer when a
2167 non-printing character is used in a format directive.
2168 * fixed several bugs in compiler checking of type declarations, i.e.
2169 violations of the Python "declarations are assertions" principle
2170 (thanks to Alexey Dejneka)
2171 * fixed several bugs in PCL's error checking (thanks to Gerd
2173 * fixed bug: printing of FILE-ERROR (thanks to Antonio
2175 * fixed bug in compilation of functions as first class values
2176 (thanks to Antonio Martinez-Shotton)
2177 * The compiler's handling TYPE-ERRORs which it can prove will
2178 inevitably happen at runtime has been cleaned up and corrected
2179 in several ways. (thanks to Alexey Dejneka)
2180 * improved argument type checking for various basic arithmetic
2181 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
2182 lavished on them in the past that they can be compiled in many
2183 ways in different special cases
2184 * fixed bug 181: compiler checks validity of user supplied type
2186 * cleaned up code flushing in optimization: Function calls which
2187 should signal errors for safety purposes (e.g. which ANSI says
2188 should signal errors when their arguments are of incorrect type)
2189 are no longer optimized away.
2190 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
2191 * incremented fasl file version number, because changes in the
2192 implementation of sequence functions like COERCE caused
2193 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
2194 in old inline expansions) to become undefined. (Actually these
2195 changes were later undone, so we might very well be binary
2196 compatible with 0.7.7 after all, but leaving the version number
2197 incremented seemed like the simplest and most conservative
2200 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
2201 * An alpha-quality port to the parisc architecture running Linux,
2202 based on the old CMUCL backend, has been made. This, even more so
2203 than the other backends, should be considered still a work in
2204 progress; known problems include that the Linux kernel in 64-bit
2205 mode does not propagate the correct sigcontext structure to
2206 userspace, and consequently SBCL on a parisc64 kernel will not
2208 * fixed bug 189: The compiler now respects NOTINLINE declarations for
2209 functions declared in FLET and LABELS. (I.e. "LET conversion" is
2210 suppressed.) Also now that the compiler is looking at declarations
2211 in the environment, it checks optimization declarations as well,
2212 and suppresses inlining when (> DEBUG SPEED).
2213 * More fixes have been made to treatment of floating point exception
2214 treatment and other Unix signals. In particular, floating point
2215 exceptions no longer cause Bus errors on the SPARC/Linux platform.
2216 * The detection and handling of control stack exhaustion (infinite
2217 or very deeply nested recursion) has changed. Stack exhaustion
2218 detection is now done by write-protecting pages at the OS level
2219 and applies at all optimization settings; when found, a
2220 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
2221 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
2222 be caught using IGNORE-ERRORS.
2223 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
2224 correctly when transforming recursive LABELS functions to LETs.
2225 (thanks to Alexey Dejneka)
2226 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
2227 that are names of constants or global variables.
2228 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
2229 alien routines with docstrings.
2230 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
2231 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
2233 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
2234 (thanks to Eric Marsden)
2235 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
2236 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
2237 * Bug fix: The SPARC backend can now compile functions involving
2238 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
2239 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
2240 to a functions expecting a PATHNAME-DESIGNATOR.
2241 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
2242 lambda lists. (thanks to Alexey Dejneka)
2243 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
2244 functions so that EXPORTing the name of the class doesn't cause
2245 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
2246 to Antonio Martinez for reporting this)
2247 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
2248 constant-folded without failing an assertion. (thanks to Einar
2249 Floystad Dorum for reporting this)
2250 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
2251 types are now dealt with more correctly by the compiler.
2252 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
2253 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
2254 behaviour analogous to RENAME-FILE. This puts its behaviour more
2255 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
2257 * The fasl file version number has changed again. (because of the
2258 bug fix involving the names of PCL MAKE-INSTANCE functions)
2260 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
2261 * bug fix: Floating point exceptions are treated much more
2262 consistently on the x86/Linux and PPC/Linux platforms.
2263 * Array initialization with :INITIAL-ELEMENT is now much faster for
2264 cases when the compiler cannot open code the array creation, but
2265 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
2266 array accesses have also seen a speed increase.
2267 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
2268 specified. (This was at the root of some bad interactions between
2269 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
2271 * bug fix: Internal error arguments for undefined functions are now
2272 computed correctly on the PPC/Linux platform.
2273 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
2274 Raymond Toy's patch for CMU CL)
2275 * Support for the Solaris 9 operating environment has been included
2276 (thanks to Daniel Merritt)
2277 * A very ugly but hopefully complete draft of the missing FFI chapter
2278 of the manual has been created by reformatting the corresponding
2279 CMU CL manual chapter into (currently very ugly and incoherent)
2280 DocBook and bringing it up to date for SBCL behavior. Thus, the
2281 manual is now essentially complete, at least by my extreme
2282 once-and-only-once standards, whereby it's acceptable to refer to
2283 the doc strings of SB-EXT functions as the primary documentation.
2284 * The fasl file version number has changed again, due to cleanup of
2285 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
2286 is no longer a static symbol.)
2288 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
2289 * SBCL now builds with OpenMCL (version 0.12) as the
2290 cross-compilation host; also, more progress has been made toward
2291 bootstrapping under CLISP.
2292 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
2294 * bug 158 fixed: The compiler can now deal with integer loop
2295 increments different from 1; fixing this turned out also to fix
2297 * bug 169 fixed: no more bogus warnings about using lexical bindings
2298 despite the presence of perfectly good SPECIAL declarations (thanks
2299 to David Lichteblau)
2300 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
2301 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
2302 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
2304 * bug 180 fixed: Method combination specifications no longer ignore
2305 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
2306 * bug fix: Structure type predicate functions now check their argument
2307 count as they should.
2308 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
2309 correctly. (thanks to Pierre Mai)
2310 * minor incompatible change: The --noprogrammer option is deprecated
2311 in favor of the new --disable-debugger option, which is very similar.
2312 (The major difference is that it takes effect at a slightly different
2313 time at startup, causing handling of errors in --sysinit and
2314 --userinit files will be affected differently.) The
2315 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
2316 been added to allow this functionality to be controlled from ordinary
2317 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
2318 maintainers, who might want to run non-interactive scripts to
2319 build SBCL cores which will later be used interactively.)
2320 * minor incompatible change: The LOAD function no longer, when given
2321 a wild pathname to load, loads all files matching that pathname.
2322 Instead, an error of type FILE-ERROR is signalled.
2324 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
2325 * bug 147 fixed: The compiler preserves its block link/count
2326 invariants more correctly now so that it doesn't crash. (thanks
2328 * Dynamic loading of object files in OpenBSD is now supported. (thanks
2330 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
2331 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
2332 no-value-for-this-character as NIL (as specified by ANSI).
2333 * HOST-NAMESTRING on physical pathnames now returns a string that is
2334 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
2335 (thanks to Christophe Rhodes)
2336 * The Alpha port handles icache flushing more correctly. (thanks to
2338 * More progress has been made toward bootstrapping under CLISP. (thanks
2339 to Christophe Rhodes)
2340 * The fasl file format has changed again, because dynamic loading
2341 on OpenBSD (which has non-ELF object files) motivated some cleanups
2342 in the way that foreign symbols are transformed and passed around.
2343 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
2344 is no longer treated as whitespace by the reader, but instead as
2345 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
2346 |A
\7fB|, instead of A as it used to.
2348 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
2349 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
2350 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
2351 * SBCL now runs on the PPC archtiecture under Linux. It actually did
2352 this as of 0.7.1.45, but was left out of the previous news section
2353 (thanks to Dan Barlow)
2354 * SBCL now runs on the Solaris operating system on SPARC architectures
2355 (thanks to Christophe Rhodes's port of the CMUCL runtime)
2356 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
2357 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
2358 * SPARC backend cleanups, allowing builds of cores optimized for V8
2359 and V9 SPARCS, and also emission of code targeted to a particular
2360 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
2362 * SBCL is closer to bootstrapping under CLISP, thanks to various
2363 fixes by Christophe Rhodes.
2364 * The fasl file format has changed again, to allow the compiler's
2365 INFO database to support symbol macros.
2366 * The user manual (in doc/) is formatted into HTML more nicely.
2367 (thanks to coreythomas)
2368 * The system is smarter about SUBTYPEP relationships, especially
2369 those involving NOT types (including types such as ATOM which are
2370 represented internally using NOT types). Thus SUBTYPEP is less
2371 likely to return (VALUES NIL NIL) in general, and in particular
2372 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
2374 * The fasl file format has changed again, because the internal
2375 representation of types now includes a new slot to support the new
2376 SUBTYPEP-of-NOT-types logic.
2377 * (not a change in the main branch of SBCL, but a related prototype
2378 which can hopefully be merged into the main branch of SBCL in the
2379 future:) Brian Spilsbury has produced a Unicode-enabled variant of
2380 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
2381 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
2382 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
2383 dispatch macro character combinations. (thanks to Alexey Dejneka)
2384 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
2385 that superficially look like logical namestrings correctly.
2386 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
2388 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
2389 default supplied-p) for &optional and &key arguments. (thanks to
2392 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
2393 * incompatible change: The compiler is now less aggressive about
2394 tail call optimization, doing it only when (> SPACE DEBUG) or
2395 (> SPEED DEBUG). (This is an incompatible change because there are
2396 programs which relied on the old CMU-CL-style behavior to optimize
2397 away their unbounded recursion which will now die of stack overflow.)
2398 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
2399 for non-GENCGC systems has been increased to 20M (since that
2400 seems much closer to the likely performance optimum for modern
2401 systems than the old 4M value was)
2402 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
2403 and *DEBUG-PRINT-LEVEL*
2404 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
2405 of CMU CL's support for SPARC, and various endianness and other
2406 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
2407 * new syntactic sugar for the Unix command line: --load foo.bar is now
2408 an alternate notation for --eval '(load "foo.bar")'.
2410 ** The system now detects stack overflow and handles it gracefully,
2411 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
2412 optimization settings. (This is a good thing in general, and
2413 its introduction in this version should be particularly timely
2414 for anyone whose code fails because of suppression of tail
2416 ** The system now hunts for the C variable "environ" in a more
2417 devious way, to avoid segfaults when the C library version differs
2418 between compile time and run time. (thanks to Christophe Rhodes)
2419 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
2420 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
2421 ** The compiler no longer issues bogus style warnings for undefined
2422 classes in the same source file as the DEFCLASSes which defined
2423 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
2425 ** fixes in CONDITION class precedence list for undefined function
2426 errors (thanks to Alexei Dejneka)
2427 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
2428 correctly. (thanks to Dan Barlow)
2429 ** portability fixes aiming at bootstrapping under CLISP (thanks
2430 to Dave McDonald and Christophe Rhodes)
2431 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
2432 ** fixes in type translation and and type inference (thanks to
2434 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
2435 ** various fixes in the new ports (thanks to Dan Barlow)
2436 * several changes related to debugging:
2437 ** suppression of tail recursion, as noted above
2438 ** stack overflow detection, as noted above
2439 ** The default implementation of TRACE has changed. :ENCAPSULATE T
2440 is now the default. (For some time encapsulation has been more
2441 reliable than the breakpoint-based :ENCAPSULATE NIL
2442 implementation, at least on X86 systems; and I just noticed that
2443 encapsulation also seems closer to the spirit of the ANSI
2446 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
2448 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
2449 up properly again. (There was a packaging bug in 0.7.0 which
2450 left their definitions in SB-SYS::LOAD-FOREIGN and
2451 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
2452 vital for most things which interface to C-level interfaces,
2453 like extensions working with sockets or databases or
2454 Perl-compatible regexes or whatever, and the need to fix
2455 this bug is the main reason that 0.7.1 was released so
2457 ** DEFGENERIC is now choosier about the methods it redefines, so that
2458 reLOADing a previously-LOADed file containing DEFGENERICs does
2459 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
2460 cycle now works as it should. (thanks to Alexey Dejneka)
2461 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
2462 fixed by Christophe Rhodes. (He actually submitted this patch
2463 months ago, and I delayed until after 0.7.0.)
2464 ** Bug 111 (internal compiler confusion about runtime checks on
2465 FUNCTION types) was fixed by Alexey Dejneka.
2466 * Some internal cleanups (getting rid of variables which aren't
2467 needed now that the byte interpreter is gone) caused the fasl
2468 file format number to change again.
2470 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
2471 * major incompatible change: The default fasl file extension, i.e. the
2472 default extension for files produced by COMPILE-FILE, has changed
2473 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
2475 ** There are many changes in the implementation of the compiler.
2476 SBCL is now essentially a compiler-only implementation of ANSI
2477 Common Lisp. EVAL still "interprets" a few special cases, but
2478 almost all the interesting cases are handled by creating
2479 a LAMBDA expression, calling COMPILE on it, then calling
2480 FUNCALL on the result.
2481 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
2482 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
2483 Since the code is newer, there might still be some new bugs
2484 (though not as many as before Martin Atzmueller's fixes:-). But
2485 the new code is substantially simpler and clearer, and hopefully
2486 any remaining bugs will be simpler, less fundamental, and more
2487 fixable then the bugs in the old code.
2488 ** The revised compiler is still a little unsteady on its feet.
2490 *** The debugging information it produces (particularly the names
2491 of FUNCTION objects) is sometimes much less useful than what
2492 the old compiler produced.
2493 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
2494 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
2495 within a MACROLET) has been temporarily weakened.
2496 ** There are new compiler optimizations for various functions:
2497 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
2498 FIND-IF-NOT, POSITION-IF-NOT, and FILL
2499 *** the math functions TRUNCATE, FLOOR, and CEILING
2500 *** the function-of-all-trades COERCE
2501 Mostly these should be transparent, but there's one
2502 potentially-annoying problem (bug 117): when the compiler
2503 inline-expands a function and does type analysis on the result,
2504 it can create control paths which have type mismatches, and
2505 when it can't prove that those control paths aren't taken,
2506 it will issue WARNINGs about the type mismatches. This is
2507 a particular problem in practice for the new sequence functions.
2508 It's not clear how this should be fixed, and for now, a
2509 workaround is given in the entry for 117 in the BUGS file.
2510 ** (Because of the interaction between the two previous items --
2511 occasional inlining problems and new inline expansions -- some
2512 of the new sequence function optimizations won't really kick in
2513 completely until debugging information, and then inlining, are
2514 straightened out in some future version.)
2515 * minor incompatible changes:
2516 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
2517 DIRECTORY no longer implicitly promotes NIL slots of its
2518 pathname argument to :WILD. In particular, when you ask for the
2519 contents of a directory (which you used to be able to do without
2520 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
2521 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
2522 ** changes in behavior that ANSI explicitly defines to be
2523 implementation dependent:
2524 *** The new compiler-only implementation still conforms with ANSI,
2525 but acts a little different than before. Besides the obvious
2526 changes in performance tradeoffs (that the cost per form passed
2527 to EVAL has gone up, and the cost per form executed by EVAL
2528 has gone down), the behavior of the system changes a little
2529 because there are no longer any interpreted function objects.
2530 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
2531 e.g. doing COMPILE on the output of interactive DEFUN is
2533 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
2535 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
2536 has changed from (:CL) to NIL.
2537 *** The CHAR-NAME of unprintable ASCII characters which, unlike
2538 e.g. #\Newline and #\Tab, don't have names specified in the
2539 ANSI Common Lisp standard, is now based on their ASCII symbolic
2540 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
2541 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
2542 are no longer used for output.
2543 ** changes in internal implementation constants:
2544 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
2545 4 million. (If your application spends a lot of time GCing and
2546 you have a lot of RAM, you might want to experiment with
2547 increasing it even more.)
2548 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
2549 However, almost all old code should still continue to work without
2550 immediate update, as SB-C-CALL is now a (deprecated) nickname
2552 ** Old operator names in the style DEF-FOO are now deprecated in
2553 favor of new corresponding names DEFINE-FOO, for consistency with
2554 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
2555 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
2556 internal symbols, but a few supported extensions like
2557 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
2558 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
2559 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
2560 etc. as you get deeper into recursive calls to the debugger
2561 command loop, instead of the old "5]", "5]]", "5]]]"
2562 sequence. (I was motivated to do this when squabbles between
2563 ILISP and SBCL left me very deeply nested in the debugger. In the
2564 short term, this change will probably provoke more ILISP/SBCL
2565 squabbles, but hopefully it will be an improvement in the long run.)
2566 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
2567 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
2568 function, since declaiming return types involving aliens is
2569 (1) annoyingly messy to do by hand and (2) vital to efficient
2570 compilation of code which calls such functions.
2571 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
2572 longer reexported by the SB-EXT package. They're solely useful
2573 for alien code, so it seems more logical that you should get
2574 them from the SB-ALIEN package, not in SB-EXT.
2575 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
2576 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
2577 features. Instead, the code that they used to control is always
2578 built into the system.
2579 * many other bug fixes
2580 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
2581 advantage of the new EVAL-WHEN stuff and to clean them up in
2582 general, and they are now more ANSI-compliant in a number of
2583 ways. Martin Atzmueller is responsible for a lot of this.
2584 ** Besides the cleanups discussed above, Martin Atzmueller fixed
2586 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
2587 *** correct ERROR type for various file operations
2588 *** some fixes for Lisp streams
2589 *** DEFMETHOD syntax checking
2590 *** changing old weird representation of debug information as
2591 strings (which, among their other deficiencies, don't transform
2592 correctly when you rename packages, and don't change their
2593 print representation when you change things like *PACKAGE*
2594 and *PRINT-LENGTH*) to symbols and lists of symbols
2595 He also made several improvements and fixed several bugs in DESCRIBE.
2596 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
2598 *** misbehavior of WRITE-STRING/WRITE-LINE
2599 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
2600 and several other LOOP problems as well
2601 *** DIRECTORY when similar filenames are present
2602 *** DEFGENERIC with :METHOD options
2603 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
2604 *** bug in the optimization of ARRAY-ELEMENT-TYPE
2605 *** argument ordering in FIND with :TEST option
2606 *** mishandled package designator argument in APROPOS-LIST
2607 *** various problems in the backquote readmacro
2608 *** a bug in APROPOS
2609 *** probably some others that I'm not describing very well here,
2610 since the CVS log documents them by reference to sbcl-devel
2611 messages, and the SourceForge archives aren't working well.:-(
2612 ** Dan Barlow improved the Alpha port (and is making progress on the
2613 PPC port, for those of you who think different).
2614 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
2615 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
2616 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
2617 continues to work on the SPARC port (for those of you in a position
2618 to look down upon our little PC-compatible boxes from a great height).
2619 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
2620 and uses that copy, rather than the current dynamic value, when
2621 it's trying to decide whether to truncate output. Thus e.g.
2622 (let ((*print-lines* 50))
2623 (pprint-logical-block (stream nil)
2625 (let ((*print-lines* 8))
2626 (print (aref possiblybigthings i) stream)))))
2627 should now truncate the logical block only at 50 lines, instead of
2628 often truncating it at 8 lines, as it did before.
2629 * The doc/cmucl/ directory, containing old CMU CL documentation
2630 from the time of the fork, is no longer part of the base system.
2631 SourceForge has shut down its anonymous FTP service, and with it
2632 my original plan for distributing the old CMU CL documentation
2633 there. For now, if you need these files you can download an old
2634 SBCL source release and extract them from it.
2635 * The fasl file version number changed again, for dozens of reasons,
2636 some of which are apparent above.
2638 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
2639 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
2640 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
2641 patch, so that DISASSEMBLE output is much nicer.
2642 * The code in the SB-PROFILE package now seems reasonably stable.
2643 I still haven't decided what the final interface should look like
2644 (I'd like PROFILE to interact cleanly with TRACE, since both
2645 facilities use function encapsulation) but if you have a need
2646 for profiling now, you can probably use it successfully with
2647 the current CMU-CL-style interface.
2648 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
2649 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
2650 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
2651 link now returns the dangling link itself, and for similar
2652 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
2653 link itself. (In these cases the old code signalled an error and
2654 looped endlessly, respectively.) Thus, DIRECTORY now works even
2655 in the presence of dangling and cyclic symbolic links.
2656 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
2657 is now a supported extension again, since the consensus on
2658 sbcl-devel was that it can be useful for ordinary development
2659 work, not just for debugging SBCL itself.
2660 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
2661 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
2662 that currently-defined functions might be redefined later with
2663 different return types.
2664 * Hash tables can be printed readably, as inspired by CMU CL code
2665 of Eric Marsden and SBCL code of Martin Atzmueller.
2666 * better error handling in CLOS method combination, thanks to
2667 Martin Atzmueller porting Pierre Mai's CMU CL patches
2668 * more overflow fixes for >16Mbyte I/O buffers
2669 * A bug in READ has been fixed, so that now a single Ctrl-D
2670 character suffices to cause end-of-file on character streams.
2671 In particular, now you only need one Ctrl-D at the command
2672 line (not two) to exit SBCL.
2673 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
2675 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
2676 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
2677 does the right thing.
2678 * The implementation of some type tests, especially for CONDITION
2679 types, is now tidier and maybe faster, due to CMU CL code
2680 originally by Douglas Crosher, ported by Martin Atzmueller.
2681 * Some math functions have been fixed, and there are new
2682 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
2683 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
2684 * (There are also some new optimizers in contrib/*-extras.lisp. Those
2685 aren't built into sbcl-0.6.13, but are a sneak preview of what's
2686 likely to be built into sbcl-0.7.0.)
2687 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
2688 CL, ported to SBCL by Martin Atzmueller)
2689 * DESCRIBE now gives more information in some cases. (Pierre Mai's
2690 patch to CMU CL, ported to SBCL by Martin Atzmueller)
2691 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
2692 * There's a new slam.sh hack to shorten the edit/compile/debug
2693 cycle for low-level changes to SBCL itself, and a new
2694 :SB-AFTER-XC-CORE target feature to control the generation of
2695 the after-xc.core file needed by slam.sh.
2696 * minor incompatible change: The ENTRY-POINTS &KEY argument to
2697 COMPILE-FILE is no longer supported, so that now every function
2698 gets an entry point, so that block compilation looks a little
2699 more like the plain vanilla ANSI section 3.2.2.3 scheme.
2700 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
2701 returns the number of bytes consed since the system started,
2702 rather than the number consed since the first time the function
2703 was called. (The new definition parallels ANSI functions like
2704 CL:GET-INTERNAL-RUN-TIME.)
2705 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
2706 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
2707 supported. Now DIRECTORY always does the abstract Common-Lisp-y
2708 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
2709 * Fasl file version numbers are now independent of the target CPU,
2710 since historically most system changes which required version
2711 number changes have affected all CPUs equally. Similarly,
2712 the byte fasl file version is now equal to the ordinary
2715 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
2716 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
2717 is no longer recognized. I apologize for this, because it was
2718 listed in SB-EXT as a supported extension, but I found that
2719 its existing behavior was poorly specified, as well as incorrectly
2720 specified, and it looked like too much of a mess to straighten it
2721 out. I have enough on my hands trying to get ANSI stuff to work..
2722 * many patches ported from CMU CL by Martin Atzmueller, with
2723 half a dozen bug fixes in pretty-printing and the debugger, and
2724 half a dozen others elsewhere
2725 * fixed bug 13: Floating point infinities are now supported again.
2726 They might still be a little bit flaky, but thanks to bug reports
2727 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
2728 as flaky as they were.
2729 * The --noprogrammer command line option is now supported. (Its
2730 behavior is slightly different in detail from what the old man
2731 page claimed it would do, but it's still appropriate under the
2732 same circumstances that the man page talks about.)
2733 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
2734 are now supported, and enabled by default. Thus, the compiler can
2735 handle many floating point and complex operations much less
2736 inefficiently. (Thus e.g. you can implement a complex FFT
2738 * The compiler now detects type mismatches between DECLAIM FTYPE
2739 and DEFUN better, and implements CHECK-TYPE more correctly, and
2740 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
2741 thanks to patches from Martin Atzmueller.
2742 * various fixes to make the cross-compiler more portable to
2743 ANSI-conforming-but-different cross-compilation hosts (notably
2744 Lispworks for Windows, following bug reports from Arthur Lemmens)
2745 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
2746 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
2747 fixed thanks to Pierre Mai's CMU CL patches.
2748 * fixes to keep the system from overflowing internal counters when
2749 it tries to use i/o buffers larger than 16M bytes
2750 * fixed bug 45a: Various internal functions required to support
2751 complex special functions have been merged from CMU CL sources.
2752 (When I was first setting up SBCL, I misunderstood a compile-time
2753 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
2754 * improved support for type intersection and union, fixing bug 12
2755 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
2756 more obscure bugs as well
2757 * some steps toward byte-compiling non-performance-critical
2758 parts of the system, courtesy of patches from Martin Atzmueller
2759 * Christophe Rhodes has made some debian packages of sbcl at
2760 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
2761 From his sbcl-devel e-mail of 2001-04-08 they're not completely
2762 stable, but are nonetheless usable. When he's ready, I'd be happy
2763 to add them to the SourceForge "File Releases" section. (And if
2764 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
2765 * new fasl file format version number (because of changes in
2766 internal representation of (OR ..) types to accommodate the new
2767 support for (AND ..) types, among other things)
2769 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
2770 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
2772 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
2774 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
2775 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
2776 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
2777 * DESCRIBE now works on structure objects again.
2778 * Most function call argument type mismatches are now handled as
2779 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
2780 know whether the function will be redefined before the call is
2781 executed. (The compiler could flag local calls with full WARNINGs,
2782 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
2783 it doesn't keep track of enough information to know whether calls
2784 are local in this sense.)
2785 * Compiler output is now more verbose, with messages truncated
2786 later than before. (There should be some supported way for users
2787 to override the default verbosity, but I haven't decided how to
2788 provide it yet, so this behavior is still controlled by the internal
2789 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
2790 src/compiler/ir1util.lisp.)
2791 * Fasl file format version numbers have increased again, because
2792 support for the Gray streams extension changes the layout of the
2793 system's STREAM objects.
2794 * The Gray subclassable streams extension now works, thanks to a
2795 patch from Martin Atzmueller.
2796 * The full LOAD-FOREIGN extension (not just the primitive
2797 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
2798 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
2799 but like most other programs, it defaults to copying the Unix
2800 environment from the original process instead of starting the
2801 new process in an empty environment.
2802 * Extensions which manipulate the Unix environment now support
2803 an :ENVIRONMENT keyword option which doesn't smash case or
2804 do other bad things. The CMU-CL-style :ENV option is retained
2805 for porting convenience.
2806 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
2807 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
2809 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
2811 * A patch from Martin Atzmueller seems to have solved the SIGINT
2812 problem, and as far as we know, signal-handling now works cleanly.
2813 (If you find any new bugs, please report them!)
2814 * The system no longer defaults Lisp source file names to types
2815 ".l", ".cl", or ".lsp", but only to ".lisp".
2816 * The compiler no longer uses special default file extensions for
2817 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
2818 seems to expect a single default extension for all compiled code,
2819 and there's no compelling reason to try to stretch the standard
2820 to allow two different extensions.) Instead, byte-compiled files
2821 default to the same extension as native-compiled files.
2822 * Fasl file format version numbers have increased again, because
2823 a rearrangement of internal implementation packages made some
2824 dumped symbols in old fasl files unreadable in new cores.
2825 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
2826 many fewer weird special cases.
2827 * Bug #17 (differing COMPILE-FILE behavior between logical and
2828 physical pathnames) has been fixed, and some related misbehavior too,
2829 thanks to a patch from Martin Atzmueller.
2830 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
2831 by Tim Moore, ported to SBCL by Martin Atzmueller.
2832 * Martin Atzmueller fixed several filesystem-related problems,
2833 including bug #36, in part by porting CMU CL patches, which were
2834 written in part by Paul Werkowski.
2835 * More compiler warnings in src/runtime/ are gone, thanks to
2836 more patches from Martin Atzmueller.
2837 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
2840 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
2842 * DESCRIBE now works on CONDITION objects.
2843 * The debugger now handles errors which arise when trying to print
2844 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
2846 * The build system now uses an additional file, customize-target-features.lisp,
2847 to allow local modifications to the target *FEATURES* list. (The point of
2848 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
2849 debugging features enabled, without having to worry about propagating it
2850 into everyone's system when I do a "cvs update".) When no
2851 customize-target-features.lisp file exists, the target *FEATURES* list
2852 should be constructed the same way as before.
2853 * fixed bugs in DEFCONSTANT ANSI-compatibility:
2854 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
2855 warn about behavior which is undefined under the ANSI spec. Note: This
2856 is specified by ANSI, but it's not very popular with programmers.
2857 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
2858 macro in the SBCL sources for an example of a workaround which you
2859 might use to make portable ANSI-standard code which does what you want.
2860 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
2861 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
2862 when it's used as a non-toplevel form. (This is required in order
2863 to implement the DEFCONSTANT-EQX macro.)
2864 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
2865 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
2866 funny things, due to bugs in the implementation of EVAL-WHEN
2867 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
2868 non-toplevel DEFCONSTANTs are uncommon.)
2869 * The core file version number and fasl file version number have been
2870 incremented, because the old noncompliant DEFCONSTANT behavior involved
2871 calling functions which no longer exist, and because I also took the
2872 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
2873 * fixed bug 1 (error handling before read-eval-print loop starts), and
2874 redid debugger restarts and related debugger commands somewhat while
2876 ** The QUIT debugger command is gone, since it did something
2877 rather different than the SB-EXT:QUIT command, and since it never
2878 worked properly outside the main toplevel read/eval/print loop.
2879 Invoking the new TOPLEVEL restart provides the same functionality.
2880 ** The GO debugger command is also gone, since you can just invoke
2881 the CONTINUE restart directly instead.
2882 ** The TOP debugger command is also gone, since it's redundant with the
2883 FRAME 0 command, and since it interfered with abbreviations for the
2885 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
2887 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
2888 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
2889 some porting work by Martin Atzmueller)
2890 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
2891 with a boatload of patches which he ported to SBCL. Now that those
2893 ** The system tries to make sure that its low-priority messages
2894 are prefixed by semicolons, to help people who like to use
2895 syntax highlighting in their ILISP buffer. (This patch
2896 was originally due to Raymond Toy.)
2897 ** The system now optimizes INTEGER-LENGTH better, thanks to more
2898 patches originally written by Raymond Toy.
2899 ** The compiler understands coercion between single-value and
2900 multiple-VALUES type expressions better, getting rid of some very
2901 weird behavior, thanks to patches originally by Robert MacLachlan
2902 and Douglas Crosher.
2903 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
2904 lambda lists, thanks to a patch originally by Pierre Mai.
2905 ** The system no longer bogusly warns about "abbreviated type
2907 ** The compiler gets less confused by inlining and RETURN-FROM,
2908 thanks to some patches originally by Tim Moore.
2909 ** The system no longer hangs when dumping circular lists to fasl
2910 files, thanks to a patch originally from Douglas Crosher.
2911 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
2912 undefined function error.
2913 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
2914 in structure slot initforms) for now, documented workaround instead:-|
2915 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
2916 * fixed bug 5: added stubs for various Gray stream functions called
2917 in the not-a-CL:STREAM case, so that even when Gray streams aren't
2918 installed, at least appropriate type errors are generated
2919 * fixed bug 8: better reporting of various PROGRAM-ERRORs
2920 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
2921 consistently in DEFMETHOD forms.
2922 * removed bug 21 from BUGS, since Martin Atzmueller points out that
2923 it doesn't seem to affect SBCL after all
2924 * The C runtime system now builds with better optimization and many
2925 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
2927 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
2929 * The system is now under CVS at SourceForge (instead of the
2930 CVS repository on my home machine).
2931 * The new signal handling code has been tweaked to treat register
2932 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
2933 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
2934 downstream, e.g. in debug-int.lisp, has implicit dependencies
2935 on the unsignedness of integer representation of machine words,
2936 and that caused the system to bomb out with infinite regress
2937 when trying to recover from type errors involving signed values,
2938 e.g. (BUTLAST '(1 2 3) -1).)
2939 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
2940 (This was one of the bugs Peter Van Eynde reported back in July.)
2941 * The system now uses code inspired by Colin Walters' O(N)
2942 implementation of MAP (from the cmucl-imp@cons.org mailing
2943 list, 2 September 2000) when it can't use a DEFTRANSFORM to
2944 inline the MAP operation, and there is more than one
2945 sequence argument to the MAP call (so that it can't just
2946 do ETYPECASE once and for all based on the type of the
2947 single sequence argument). (The old non-inline implementation
2948 of the general M-argument sequence-of-length-N case required
2949 O(M*N*N) time when any of the sequence arguments were LISTs.)
2950 * The QUIT :UNIX-CODE keyword argument has been renamed to
2951 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
2952 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
2954 * Raymond Wiker's patches to port dynamic loading from Linux to
2955 FreeBSD have been added.
2956 * The BUGS file is now more nearly up to date, thanks in large part
2957 to Martin Atzmueller's review of it.
2958 * The debugger now flushes standard output streams before it begins
2959 its output ("debugger invoked" and so forth).
2960 * The core version number and fasl file version number have both
2961 been incremented, because of incompatible changes in the layout
2963 * FINISH-OUTPUT is now called more consistently on QUIT. (It
2964 used to not be called for a saved Lisp image.)
2965 * Martin Atzmueller's version of a patch to fix a compiler crash,
2966 as posted on sbcl-devel 13 September 2000, has been installed.
2967 * Instead of installing Martin Atzmueller's patch for the
2968 compiler transform for SUBSEQ, I deleted the compiler transform,
2969 and transforms for some similar consing operations.
2970 * A bug in signal handling which kept TRACE from working on OpenBSD
2972 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
2973 away by constant folding
2974 * The system now defines its address space constants in one place
2975 (in the Lisp sources), and propagates them automatically elsewhere
2976 (through GENESIS and the sbcl.h file). Therefore, patching the
2977 address map is less unnecessarily tedious and error-prone. The
2978 Lisp names of address space constants have also been systematized.
2979 * CVS tags like dollar-Header-dollar have been removed from
2980 the sources, because they have never saved me trouble and
2981 they've been source of trouble working with patches and other
2982 diff-related operations.
2983 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
2984 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
2986 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
2988 * The system has been ported to OpenBSD.
2989 * The system now compiles with a simple "sh make.sh" on the systems
2990 that it's supported on. I.e., now you no longer need to tweak
2991 text in the target-features.lisp-expr and symlinks in src/runtime/
2992 by hand, the make.sh takes care of it for you.
2993 * The system is no longer so grossly inefficient when compiling code
2994 involving vectors implemented as general (not simple) vectors (VECTOR T),
2995 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
2996 which dares to use the various sequence functions on non-simple
2997 vectors, takes less of a performance hit.
2998 * There is now a primitive type predicate VECTOR-T-P
2999 to test for the (VECTOR T) type, so that e.g.
3000 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
3001 can now be compiled with some semblance of efficiency. (The old code
3002 turned the type declaration into a full call to %TYPEP at runtime!)
3003 * AREF on (VECTOR T) is still not fast, since it's still compiled
3004 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
3005 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
3006 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
3007 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
3008 without making full calls to SUBTYPEP at runtime.
3009 (Some analogous efficiency issues for non-simple vectors specialized to
3010 element types other than T, or for non-simple multidimensional arrays,
3011 have not been addressed. They could almost certainly be handled the
3012 same way if anyone is motivated to do so.)
3013 * The changes in array handling break binary compatibility, so
3014 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
3015 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
3016 instead of (VALUES T T).
3017 * By following the instructions that Dan Barlow posted to sbcl-devel
3018 on 2 July 2000, I was able to enable primitive dynamic object
3019 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
3020 functionality is not implemented (since it calls ld to resolve
3021 library references automatically, requiring RUN-PROGRAM for its
3022 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
3023 try to resolve library references) is now supported.
3024 * The system now flushes the standard output streams when it terminates,
3025 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
3026 them at several other probably-convenient times, e.g. in each pass of
3027 the toplevel read-eval-print loop, and after evaluating a form given
3028 as an "--eval" command-line option. (These changes were motivated by a
3029 discussion of stream flushing issues on cmucl-imp in August 2000.)
3030 * The source transform for TYPEP of array types no longer assumes
3031 that an array whose element type is a not-yet-defined type
3032 is implemented as an array of T, but instead punts, so that the
3033 type will be interpreted at runtime.
3034 * There is now some support for cross-compiling in make.sh: each of
3035 the phases of make.sh has its own script. (This should be transparent
3036 to people doing ordinary, non-cross-compile builds.)
3037 * Since my laptop doesn't have hundreds of megabytes of memory like
3038 my desktop machine, I became more motivated to do some items on
3039 my to-do list in order to reduce the size of the system a little:
3040 ** Arrange for various needed-only-at-cold-init things to be
3041 uninterned after cold init. To support this, those things have
3042 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
3043 symbols with such names are now uninterned after cold init).
3044 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
3045 fasl files for cold load.
3046 ** Remove the old compiler structure pooling code (which used to
3047 be conditional on the target feature :SB-ALLOC) completely.
3048 ** Redo the representation of some data in cold init to be more compact.
3049 (I also looked into supporting byte compiled code at bootstrap time,
3050 which would probably reduce the size of the system a lot, but that
3051 looked too complicated, so I punted for now.)
3052 * The maximum signal nesting depth in the src/runtime/ support code has
3053 been reduced from 4096 to 256. (I don't know any reason for the very
3054 large old value. If the new smaller value turns out to break something,
3055 I'll probably just bump it back up.)
3056 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
3058 * Many, many bugs reported by Peter Van Eynde have been added to
3059 the BUGS list; some have even been fixed.
3060 * While enabling dynamic object file loading, I tried to make the
3061 code easier to understand, renaming various functions and variables
3062 with less ambiguous names, and changing some function calling
3063 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
3064 * While trying to figure out how to do the OpenBSD port, I tried to
3065 clean up some of the code in src/runtime/. In particular, I dropped
3066 support for non-POSIX signal handling, added various comments,
3067 tweaked the code to reduce the number of compilation warnings, and
3068 renamed some files to increase consistency.
3069 * To support the new automatic configuration functionality in make.sh,
3070 the source file target-features.lisp-expr has been replaced with the
3071 source file base-target-features.lisp-expr and the machine-generated
3072 file local-target-features.lisp-expr.
3073 * fixed a stupid quoting error in make.sh so that using CMU CL
3074 "lisp -batch" as cross-compilation host works again
3076 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
3078 * DESCRIBE no longer tries to call itself recursively to describe
3079 bound/fbound values, so that it no longer fails on symbols which are
3080 bound to themselves (like keywords, T, and NIL).
3081 * DESCRIBE now works on generic functions.
3082 * The printer now prints less-screwed-up representations of closures
3083 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
3084 * A private symbol is used instead of the :EMPTY keyword previously
3085 used to mark empty slots in hash tables. Thus
3086 (DEFVAR *HT* (MAKE-HASH-TABLE))
3087 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
3088 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
3089 now does what ANSI says that it should. (You can still get
3090 similar noncompliant behavior if bang on the hash table
3091 implementation with all the symbols you get back from
3092 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
3093 This breaks binary compatibility, since tests for equality to
3094 :EMPTY are wired into things like the macroexpansion of
3095 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
3097 * There's now a minimal placeholder implementation for CL:STEP,
3098 as required by ANSI.
3099 * An obscure bug in the interaction of the normal compiler, the byte
3100 compiler, inlining, and structure predicates has been patched
3101 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
3102 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
3104 * Missing ordinary arguments in a macro call are now detected even
3105 when the macro lambda list contains &KEY or &REST.
3106 * The debugger no longer complains about encountering the top of the
3107 stack when you type "FRAME 0" to explicitly instruct it to go to
3108 the top of the stack. And it now prints the frame you request even
3109 if it's the current frame (instead of saying "You are here.").
3110 * As specified by ANSI, the system now always prints keywords
3111 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
3113 * The default initial SIZE of HASH-TABLEs is now smaller.
3114 * Type information from CLOS class dispatch is now propagated
3115 into DEFMETHOD bodies, so that e.g.
3116 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3118 is now basically equivalent to
3119 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3120 (DECLARE (TYPE SINGLE-FLOAT X))
3122 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
3123 operation, without having to do run-time type dispatch.
3124 * The macroexpansion of DEFMETHOD has been tweaked so that it has
3125 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
3126 * Since I don't seem to be making big file reorganizations very often
3127 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
3128 is overflowing my ability to conveniently back them up), I've finally
3129 checked the system into CVS. (The CVS repository is on my home system,
3130 not at SourceForge -- putting it on SourceForge might come later.)
3131 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
3132 high-level GC-NOTIFY-FOO functions send their output. (There's
3133 still very little control of where low-level verbose GC functions
3134 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
3135 less than it used to -- the GC-NOTIFY-FOO functions are now under
3136 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
3137 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
3138 in only one place in the source code, and propagates it automatically
3139 everywhere that it's needed. Thus e.g. when I bump the version from
3140 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
3141 * The C source files now include boilerplate legalese and documentation
3142 at the head of each file (just as the Lisp source files already did).
3143 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
3144 to his page will be replaced with a link to his new CLiki service.
3146 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
3148 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
3149 * The build process now looks for GNU make under the default name "gmake",
3150 instead of "make" as it used to. If GNU make is not available as "gmake"
3151 on your system, you can change this default behavior by setting the
3152 GNUMAKE environment variable.
3153 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
3154 can build without error under CMU CL.
3156 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
3158 * There is now a partial SBCL user manual (with some new text and some
3159 text cribbed from the CMU CL manual).
3160 * The beginnings of a profiler have been added (starting with the
3161 CMU CL profiler and simplifying and cleaning up). Eventually the
3162 main interface should be through the TRACE macro, but for now,
3163 it's still accessed through vaguely CMU-CL-style functions and macros
3164 exported from the package SB-PROFILE.
3165 * Some problems left over from porting CMU CL to the new
3166 cross-compilation bootstrap process have been cleaned up:
3167 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
3168 instead of SB!XC:DEFMACRO, compounded by an oversight on my
3169 part when getting rid of the compiler *BACKEND* stuff.)
3170 ** The value of *NULL-TYPE* was screwed up, because it was
3171 being initialized before the type system knew the final
3172 definition of the 'NULL type. This screwed up several key
3173 optimizations in the compiler, causing inefficiency in all sorts
3174 of places. (I found it because I wanted to understand why
3175 GET-INTERNAL-RUN-TIME was consing.)
3176 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
3177 PROCLAIM FTYPE information. Unfortunately this broke binary
3178 compatibility again, since now the forms output by DEFGENERIC
3179 to refer to functions which didn't exist in 0.6.3.
3180 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
3181 can use the new (as of 0.6.3) transform for SOME into MAP into
3183 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
3184 values to (INTEGER 0 1000000), so that the time code will no longer
3185 occasionally get blown up by Linux returning 1000000 microseconds
3186 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
3187 its output conform to the ANSI spec. (Alas, this makes its output
3188 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
3189 * A full call to MAP NIL with a single sequence argument no longer conses.
3190 * fixes to problems pointed out by Martin Atzmueller:
3191 * The manual page no longer talks about multiprocessing as though
3192 it were currently supported.
3193 * The ILISP support patches have been removed from the distribution,
3194 because as of version 5.10.1, ILISP now supports SBCL without us
3195 having to maintain patches.
3196 * added a modified version of Raymond Toy's recent CMU CL patch for
3197 EQUALP comparison of HASH-TABLE
3199 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
3201 * The system still can't cross-compile itself with
3202 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
3203 entails), but at least it can compile more of itself that way
3204 than it used to be able to, and various buglets which were uncovered
3205 by trying to cross-compile itself that way have now been fixed.
3206 * This release breaks binary compatibility again. This time
3207 at least I've incremented the FASL file format version to 2, so that the
3208 problem can be detected reliably instead of just causing weird errors.
3209 * various new style warnings:
3210 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
3211 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
3212 ** using the lexical binding of a variable named in the *FOO* style
3213 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
3214 as specified by ANSI.
3215 * *RANDOM-STATE* is no longer automatically initialized from
3216 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
3217 default behavior of the system is to repeat its behavior every time
3218 it's run. If you'd like to change this behavior, you can always
3219 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
3220 old convention there was no comparably easy way to get the system to
3221 repeat its behavior every time it was run.
3222 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
3223 FTYPE declarations has been removed, in favor of their ANSI
3224 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
3225 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
3226 the types of their sequence arguments can be determined at compile time.
3227 This is done through a new open code expansion for MAP which eliminates
3228 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
3229 argument types can be determined at compile time.
3230 * The optimizer now transforms COERCE into an identity operation when it
3231 can prove that the coerced object is already of the correct type. (This
3232 can be a win for machine generated code, including the output of other
3233 optimization transforms, such as the MAP transform above.)
3234 * Credit information has been moved from source file headers into CREDITS.
3235 * Source file headers have been made more standard.
3236 * The CASE macro now compiles without complaining even when it has
3239 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
3241 * (Note that the way that the PCL macroexpansions were rewritten
3242 to accommodate the change in DEFGENERIC below breaks binary
3243 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
3244 not run under sbcl-0.6.2. Once we get out of alpha releases,
3245 i.e. hit release 1.0.0, we'll probably try to maintain binary
3246 compatibility between maintenance releases, e.g. between sbcl-1.4.3
3247 and sbcl-1.4.4. Until then, however, it might be fairly common
3248 for maintenance releases to break binary compatibility.)
3249 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
3251 * The reporting of unhandled signals has been changed to print some
3252 explanatory text as well as the report form. (Previously only
3253 the report form was printed.)
3254 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
3255 it defines, so that the compiler no longer issues undefined function
3256 warnings for compiled-but-not-yet-loaded generic functions.
3257 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
3258 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
3259 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
3260 section of the ANSI spec are supported.
3261 * The "" nickname for the "KEYWORD" package has been removed.
3262 The reader still handles symbol tokens which begin with a package marker
3263 as keywords, but it doesn't expose its mechanism for doing so in the
3264 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
3265 * The system now issues STYLE-WARNINGs for contradictory TYPE
3266 proclamations. (Warnings for contradictory FTYPE proclamations would
3267 be nice too, but those can't be done usefully unless the type system
3268 is made smarter about FUNCTION types.)
3269 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
3270 systematized, so that "*target-*.lisp is supposed to exist only on the
3271 target and imply that there's a related file which exists on the
3272 host, and *host-*.lisp is supposed to exist only on the host and imply
3273 that there's a related file which exists on the target. This involves a
3274 lot of renaming. Hopefully the acute confusion caused by the renaming
3275 will be justified by the reduction in chronic confusion..
3276 ** runtime-type.lisp -> early-target-type.lisp
3277 ** target-type.lisp -> late-target-type.lisp
3278 ** early-host-format.lisp -> early-format.lisp
3279 ** late-host-format.lisp -> late-format.lisp
3280 ** host-error.lisp -> misc-error.lisp
3281 ** early-error.lisp -> early-target-error.lisp
3282 ** late-error.lisp -> late-target-error.lisp
3283 ** host-defboot.lisp -> early-defboot.lisp
3284 ** code/misc.lisp -> code/target-misc.lisp
3285 ** code/host-misc.lisp -> code/misc.lisp
3286 ** code/numbers.lisp -> code/target-numbers.lisp
3287 ** code/early-numbers.lisp -> numbers.lisp
3288 ** early-host-type.lisp -> early-type.lisp
3289 ** late-host-type.lisp -> late-type.lisp
3290 ** host-typep.lisp -> typep.lisp
3291 ** load.lisp -> target-load.lisp
3292 ** host-load.lisp -> load.lisp
3293 ** host-disassem.lisp -> disassem.lisp
3294 ** host-insts.lisp -> insts.lisp
3295 ** byte-comp.lisp -> target-byte-comp.lisp
3296 ** host-byte-comp.lisp -> byte-comp.lisp
3297 ** host-signal.lisp -> signal.lisp
3298 ** host-defstruct.lisp -> defstruct.lisp
3299 ** late-target-type.lisp -> deftypes-for-target.lisp
3300 Furthermore, several other previously target-only files foo.lisp (e.g.
3301 hash-table.lisp and random.lisp) have been split into a target-and-host
3302 foo.lisp file and a target-only target-foo.lisp file, with their key type
3303 definitions in the target-and-host part, so that the cross-compiler will
3304 know more about target types.
3305 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
3306 gone away. In their place are various *BACKEND-FOO* variables
3307 corresponding to the slots of the old structure.
3308 * A bug which caused the SB-COLD bootstrap-time package to be propagated
3309 into the target SBCL has been fixed.
3310 * The chill.lisp system for loading cold code into a running SBCL
3312 * Support for the CMU CL "scavenger hook" extension has been removed.
3313 (It was undocumented and unused in the CMU CL sources that SBCL was
3314 derived from, and stale in sbcl-0.6.1.)
3315 * Various errors in the cross-compiler type system were detected
3316 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
3317 (enabling various consistency checks). Many of them were fixed,
3318 but some hard problems remain, so the compiler is back to
3319 running without *TYPE-SYSTEM-INITIALIZED* for now.
3320 * As part of the cross-compiler type system cleanup, I implemented
3321 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
3322 * I have started adding UNCROSS calls throughout the type system
3323 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
3324 on cross-compiler input files will be able to go away, and various
3326 * CONSTANTP now returns true for quoted forms (as explicitly required
3329 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
3331 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
3332 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
3333 (MAKE-LIST -1). In the longer term, it ought to become true
3334 that declarations are assertions even at SAFETY 1. For now, it's not
3335 quite true even at SAFETY 3, but it's at least more nearly true..
3336 (Note that this change seems to increases the size of the system by
3337 O(5%) and to decrease the speed of the compiler by 20% or more.)
3338 * changed ALIEN printing to be much more abbreviated, as a short-term fix
3339 for the problem of printing dozens of lines of distracting information
3340 about low-level system machinery as part of the top stack frame
3341 on entry to the debugger when an undefined function was called.
3342 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
3343 is not reset to COMMON-LISP-USER.
3344 * Compilation of stuff related to dyncount.lisp has been made conditional
3345 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
3346 smaller. The various dyncount-related symbols have been moved into
3347 a new "SB-DYNCOUNT" package.
3348 * tty-inspect.lisp has been renamed to inspect.lisp.
3349 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
3350 feature has gone away. (When we eventually port to other flavors of
3351 libc and/or Unix, we'll try to make the differences between flavors
3352 invisible at the user level.)
3353 * Various other *FEATURES* tags, and/or their associated conditionals,
3354 have been removed if obsolescent, or given better documentation, or
3355 sometimes given more-mnemonic names.
3357 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
3359 * tidied up "make.sh" script
3360 * tidied up system directory structure
3361 * better "clean.sh" behavior
3362 * added doc/FOR-CMUCL-DEVELOPERS
3363 * many many small tweaks to output format, e.g. removing possibly-confusing
3364 trailing #\. character in DESCRIBE-INSTANCE
3365 * (EQUALP #\A 'A) no longer signals an error.
3366 * new hashing code, including EQUALP hashing
3367 * tidied up Lisp initialization and toplevel
3368 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
3369 * command line argument processing
3370 * added POSIX-GETENV function to deal with Unix-ish environment variables
3371 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
3372 terminating SBCL on EOF
3373 * non-verbose GC by default
3374 * There is no more "sbcl" shell script; the sbcl file is now the C
3375 runtime executable (just like CMU CL).
3376 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
3378 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
3379 symbol and package information as Lisp native symbol and package objects
3380 instead of strings naming symbols and strings naming packages. This way,
3381 whenever packages are renamed (as in warm init), debug information is
3382 transformed along with everything else.
3383 * tweaked the optimization policy declarations which control the building
3384 of SBCL itself. Now, among other things, the system no longer saves
3385 source location debugging information. (This helps two problems at once
3386 by reducing SBCL size and by keeping SBCL from trying to look for its
3387 sources -- which may not exist -- when reporting errors.)
3388 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
3389 debugging and testing purposes
3390 * cleaned up printing, making the printer call PRINT-OBJECT for
3391 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
3392 methods, giving nearly-ANSI behavior
3393 * converted almost all special variables to use *FOO* naming convention
3394 * deleted PARSE-TIME functionality, since it can be done portably
3395 * moved some files out of cold init into warm init
3396 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
3398 * regularized formatting of source files
3399 * added an install.sh script
3400 * fixed ridiculous memory usage of cross-compiler by making
3401 compiler/alloc.lisp not try to do pooling unless it can hook
3402 itself into the GC of the cross-compilation host. Now the system
3403 builds nicely on my old laptop.
3404 * added :SB-ALLOC in target-features.lisp-expr
3405 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
3406 was not implemented)
3407 * re-did condition handling and note reporting in the compiler. Notes
3408 are no longer handled by signalling conditions. Style warnings
3409 and warnings are handled more correctly and reported in such a way
3410 that it's easy to find one or the other in your output (so that you
3411 can e.g. figure out which of many problems caused COMPILE-FILE to
3413 * changed the severity of several compiler warnings from full WARNING
3414 to STYLE-WARNING in order to conform with the ANSI spec; also changed
3415 compiler note reporting so that it doesn't use the condition system
3416 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
3417 COMPILE-FILE command)
3418 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
3419 function. As a consequence, START-BLOCK and END-BLOCK declarations are
3420 no longer supported, since their implementation was deeply intertwingled
3421 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
3422 * removed bogus "support" for compiler macros named (SETF FOO), and
3423 removed the compiler macro for SETF INFO (but only after making a fool
3424 of myself on the cmucl-imp mailing list by posting a bogus patch for
3425 DEFINE-COMPILER-MACRO..)
3426 * Compiled files containing forms which have side effects on the Lisp
3427 reader (such as DEFPACKAGE forms) are now handled more correctly.
3428 (Compiler queuing of top level lambdas has been suppressed by setting
3429 *TOP-LEVEL-LAMBDA-MAX* to 0. )
3430 * deleted various currently-unused source files, e.g. gengc.lisp. They
3431 may be added back at some point e.g. when porting to other architectures,
3432 but until they are it's distracting to distribute them and to try to
3434 * deleted "UNCROSS couldn't recurse through.." style warnings, since
3435 there were so many of them they're just distractions, and UNCROSS is
3436 known to be able to handle the current sources
3437 * moved PROFILE functionality into TRACE, so that it will be clear
3438 how the wrapping and unwrapping of functions when you profile them
3439 interacts with the wrapping and unwrapping of functions when you
3440 trace them. (Actually, the functionality isn't there yet, but at least
3441 the interface specification is there. Hopefully, the functionality will
3442 arrive with some future maintenance release.)
3443 * removed host-oops.lisp
3444 * changed signature of QUIT function to allow UNIX-CODE argument
3445 * fixed READ-SEQUENCE bug
3446 * tweaked verbose GC output so that it looks more like the progress
3447 output that ANSI specifies for functions like LOAD
3448 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
3449 * added <http://sbcl.sourceforge.com> to the banner information printed by